Modify YOLOv7 Infer file and the external model namespace (#87)

* first commit for yolov7

* pybind for yolov7

* CPP README.md

* CPP README.md

* modified yolov7.cc

* README.md

* python file modify

* delete license in fastdeploy/

* repush the conflict part

* README.md modified

* README.md modified

* file path modified

* file path modified

* file path modified

* file path modified

* file path modified

* README modified

* README modified

* move some helpers to private

* add examples for yolov7

* api.md modified

* api.md modified

* api.md modified

* YOLOv7

* yolov7 release link

* yolov7 release link

* yolov7 release link

* copyright

* change some helpers to private

* change variables to const and fix documents.

* gitignore

* Transfer some funtions to private member of class

* Transfer some funtions to private member of class

* Merge from develop (#9)

* Fix compile problem in different python version (#26)

* fix some usage problem in linux

* Fix compile problem

Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>

* Add PaddleDetetion/PPYOLOE model support (#22)

* add ppdet/ppyoloe

* Add demo code and documents

* add convert processor to vision (#27)

* update .gitignore

* Added checking for cmake include dir

* fixed missing trt_backend option bug when init from trt

* remove un-need data layout and add pre-check for dtype

* changed RGB2BRG to BGR2RGB in ppcls model

* add model_zoo yolov6 c++/python demo

* fixed CMakeLists.txt typos

* update yolov6 cpp/README.md

* add yolox c++/pybind and model_zoo demo

* move some helpers to private

* fixed CMakeLists.txt typos

* add normalize with alpha and beta

* add version notes for yolov5/yolov6/yolox

* add copyright to yolov5.cc

* revert normalize

* fixed some bugs in yolox

* fixed examples/CMakeLists.txt to avoid conflicts

* add convert processor to vision

* format examples/CMakeLists summary

* Fix bug while the inference result is empty with YOLOv5 (#29)

* Add multi-label function for yolov5

* Update README.md

Update doc

* Update fastdeploy_runtime.cc

fix variable option.trt_max_shape wrong name

* Update runtime_option.md

Update resnet model dynamic shape setting name from images to x

* Fix bug when inference result boxes are empty

* Delete detection.py

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>

* first commit for yolor

* for merge

* Develop (#11)

* Fix compile problem in different python version (#26)

* fix some usage problem in linux

* Fix compile problem

Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>

* Add PaddleDetetion/PPYOLOE model support (#22)

* add ppdet/ppyoloe

* Add demo code and documents

* add convert processor to vision (#27)

* update .gitignore

* Added checking for cmake include dir

* fixed missing trt_backend option bug when init from trt

* remove un-need data layout and add pre-check for dtype

* changed RGB2BRG to BGR2RGB in ppcls model

* add model_zoo yolov6 c++/python demo

* fixed CMakeLists.txt typos

* update yolov6 cpp/README.md

* add yolox c++/pybind and model_zoo demo

* move some helpers to private

* fixed CMakeLists.txt typos

* add normalize with alpha and beta

* add version notes for yolov5/yolov6/yolox

* add copyright to yolov5.cc

* revert normalize

* fixed some bugs in yolox

* fixed examples/CMakeLists.txt to avoid conflicts

* add convert processor to vision

* format examples/CMakeLists summary

* Fix bug while the inference result is empty with YOLOv5 (#29)

* Add multi-label function for yolov5

* Update README.md

Update doc

* Update fastdeploy_runtime.cc

fix variable option.trt_max_shape wrong name

* Update runtime_option.md

Update resnet model dynamic shape setting name from images to x

* Fix bug when inference result boxes are empty

* Delete detection.py

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>

* Yolor (#16)

* Develop (#11) (#12)

* Fix compile problem in different python version (#26)

* fix some usage problem in linux

* Fix compile problem

Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>

* Add PaddleDetetion/PPYOLOE model support (#22)

* add ppdet/ppyoloe

* Add demo code and documents

* add convert processor to vision (#27)

* update .gitignore

* Added checking for cmake include dir

* fixed missing trt_backend option bug when init from trt

* remove un-need data layout and add pre-check for dtype

* changed RGB2BRG to BGR2RGB in ppcls model

* add model_zoo yolov6 c++/python demo

* fixed CMakeLists.txt typos

* update yolov6 cpp/README.md

* add yolox c++/pybind and model_zoo demo

* move some helpers to private

* fixed CMakeLists.txt typos

* add normalize with alpha and beta

* add version notes for yolov5/yolov6/yolox

* add copyright to yolov5.cc

* revert normalize

* fixed some bugs in yolox

* fixed examples/CMakeLists.txt to avoid conflicts

* add convert processor to vision

* format examples/CMakeLists summary

* Fix bug while the inference result is empty with YOLOv5 (#29)

* Add multi-label function for yolov5

* Update README.md

Update doc

* Update fastdeploy_runtime.cc

fix variable option.trt_max_shape wrong name

* Update runtime_option.md

Update resnet model dynamic shape setting name from images to x

* Fix bug when inference result boxes are empty

* Delete detection.py

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>

* Develop (#13)

* Fix compile problem in different python version (#26)

* fix some usage problem in linux

* Fix compile problem

Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>

* Add PaddleDetetion/PPYOLOE model support (#22)

* add ppdet/ppyoloe

* Add demo code and documents

* add convert processor to vision (#27)

* update .gitignore

* Added checking for cmake include dir

* fixed missing trt_backend option bug when init from trt

* remove un-need data layout and add pre-check for dtype

* changed RGB2BRG to BGR2RGB in ppcls model

* add model_zoo yolov6 c++/python demo

* fixed CMakeLists.txt typos

* update yolov6 cpp/README.md

* add yolox c++/pybind and model_zoo demo

* move some helpers to private

* fixed CMakeLists.txt typos

* add normalize with alpha and beta

* add version notes for yolov5/yolov6/yolox

* add copyright to yolov5.cc

* revert normalize

* fixed some bugs in yolox

* fixed examples/CMakeLists.txt to avoid conflicts

* add convert processor to vision

* format examples/CMakeLists summary

* Fix bug while the inference result is empty with YOLOv5 (#29)

* Add multi-label function for yolov5

* Update README.md

Update doc

* Update fastdeploy_runtime.cc

fix variable option.trt_max_shape wrong name

* Update runtime_option.md

Update resnet model dynamic shape setting name from images to x

* Fix bug when inference result boxes are empty

* Delete detection.py

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>

* documents

* documents

* documents

* documents

* documents

* documents

* documents

* documents

* documents

* documents

* documents

* documents

* Develop (#14)

* Fix compile problem in different python version (#26)

* fix some usage problem in linux

* Fix compile problem

Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>

* Add PaddleDetetion/PPYOLOE model support (#22)

* add ppdet/ppyoloe

* Add demo code and documents

* add convert processor to vision (#27)

* update .gitignore

* Added checking for cmake include dir

* fixed missing trt_backend option bug when init from trt

* remove un-need data layout and add pre-check for dtype

* changed RGB2BRG to BGR2RGB in ppcls model

* add model_zoo yolov6 c++/python demo

* fixed CMakeLists.txt typos

* update yolov6 cpp/README.md

* add yolox c++/pybind and model_zoo demo

* move some helpers to private

* fixed CMakeLists.txt typos

* add normalize with alpha and beta

* add version notes for yolov5/yolov6/yolox

* add copyright to yolov5.cc

* revert normalize

* fixed some bugs in yolox

* fixed examples/CMakeLists.txt to avoid conflicts

* add convert processor to vision

* format examples/CMakeLists summary

* Fix bug while the inference result is empty with YOLOv5 (#29)

* Add multi-label function for yolov5

* Update README.md

Update doc

* Update fastdeploy_runtime.cc

fix variable option.trt_max_shape wrong name

* Update runtime_option.md

Update resnet model dynamic shape setting name from images to x

* Fix bug when inference result boxes are empty

* Delete detection.py

Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>
Co-authored-by: Jason <928090362@qq.com>

* add is_dynamic for YOLO series (#22)

* modify namespace for yolov5lite

* modify namespace for external detection models

* modify namespace for yolov5lite

* modify namespace for face models

* py namespace for external model

* fix compile error and infer file

* change paddleyolox to yolox

* delete space in cmakelist

* check code specification

Co-authored-by: Jason <jiangjiajun@baidu.com>
Co-authored-by: root <root@bjyz-sys-gpu-kongming3.bjyz.baidu.com>
Co-authored-by: DefTruth <31974251+DefTruth@users.noreply.github.com>
Co-authored-by: huangjianhui <852142024@qq.com>
Co-authored-by: Jason <928090362@qq.com>

csrcs/fastdeploy/vision.h

@@ -15,29 +15,27 @@
 
 #include "fastdeploy/core/config.h"
 #ifdef ENABLE_VISION
-#include "fastdeploy/vision/biubug6/retinaface.h"
+#include "fastdeploy/vision/detection/contrib/nanodet_plus.h"
-#include "fastdeploy/vision/deepcam/yolov5face.h"
-#include "fastdeploy/vision/deepinsight/arcface.h"
-#include "fastdeploy/vision/deepinsight/cosface.h"
-#include "fastdeploy/vision/deepinsight/insightface_rec.h"
-#include "fastdeploy/vision/deepinsight/partial_fc.h"
-#include "fastdeploy/vision/deepinsight/scrfd.h"
-#include "fastdeploy/vision/deepinsight/vpl.h"
-
 #include "fastdeploy/vision/detection/contrib/scaledyolov4.h"
 #include "fastdeploy/vision/detection/contrib/yolor.h"
+#include "fastdeploy/vision/detection/contrib/yolov5.h"
+#include "fastdeploy/vision/detection/contrib/yolov5lite.h"
+#include "fastdeploy/vision/detection/contrib/yolov6.h"
 #include "fastdeploy/vision/detection/contrib/yolov7.h"
-#include "fastdeploy/vision/detection/ppdet/model.h"
+#include "fastdeploy/vision/detection/contrib/yolox.h"
-
+#include "fastdeploy/vision/facedet/contrib/retinaface.h"
-#include "fastdeploy/vision/linzaer/ultraface.h"
+#include "fastdeploy/vision/facedet/contrib/scrfd.h"
-#include "fastdeploy/vision/megvii/yolox.h"
+#include "fastdeploy/vision/facedet/contrib/ultraface.h"
-#include "fastdeploy/vision/meituan/yolov6.h"
+#include "fastdeploy/vision/facedet/contrib/yolov5face.h"
+#include "fastdeploy/vision/faceid/contrib/arcface.h"
+#include "fastdeploy/vision/faceid/contrib/cosface.h"
+#include "fastdeploy/vision/faceid/contrib/insightface_rec.h"
+#include "fastdeploy/vision/faceid/contrib/partial_fc.h"
+#include "fastdeploy/vision/faceid/contrib/vpl.h"
+#include "fastdeploy/vision/matting/contrib/modnet.h"
 #include "fastdeploy/vision/ppcls/model.h"
-#include "fastdeploy/vision/ppogg/yolov5lite.h"
+#include "fastdeploy/vision/detection/ppdet/model.h"
 #include "fastdeploy/vision/ppseg/model.h"
-#include "fastdeploy/vision/rangilyu/nanodet_plus.h"
-#include "fastdeploy/vision/ultralytics/yolov5.h"
-#include "fastdeploy/vision/zhkkke/modnet.h"
 #endif
 
 #include "fastdeploy/vision/visualize/visualize.h"

csrcs/fastdeploy/vision/deepinsight/deepinsight_pybind.cc

@@ -1,144 +0,0 @@
-// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-#include "fastdeploy/pybind/main.h"
-
-namespace fastdeploy {
-void BindDeepInsight(pybind11::module& m) {
-  auto deepinsight_module =
-      m.def_submodule("deepinsight", "https://github.com/deepinsight");
-  // Bind SCRFD
-  pybind11::class_<vision::deepinsight::SCRFD, FastDeployModel>(
-      deepinsight_module, "SCRFD")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::deepinsight::SCRFD& self, pybind11::array& data,
-              float conf_threshold, float nms_iou_threshold) {
-             auto mat = PyArrayToCvMat(data);
-             vision::FaceDetectionResult res;
-             self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
-             return res;
-           })
-      .def_readwrite("size", &vision::deepinsight::SCRFD::size)
-      .def_readwrite("padding_value",
-                     &vision::deepinsight::SCRFD::padding_value)
-      .def_readwrite("is_mini_pad", &vision::deepinsight::SCRFD::is_mini_pad)
-      .def_readwrite("is_no_pad", &vision::deepinsight::SCRFD::is_no_pad)
-      .def_readwrite("is_scale_up", &vision::deepinsight::SCRFD::is_scale_up)
-      .def_readwrite("stride", &vision::deepinsight::SCRFD::stride)
-      .def_readwrite("use_kps", &vision::deepinsight::SCRFD::use_kps)
-      .def_readwrite("max_nms", &vision::deepinsight::SCRFD::max_nms)
-      .def_readwrite("downsample_strides",
-                     &vision::deepinsight::SCRFD::downsample_strides)
-      .def_readwrite("num_anchors", &vision::deepinsight::SCRFD::num_anchors)
-      .def_readwrite("landmarks_per_face",
-                     &vision::deepinsight::SCRFD::landmarks_per_face);
-  // Bind InsightFaceRecognitionModel
-  pybind11::class_<vision::deepinsight::InsightFaceRecognitionModel,
-                   FastDeployModel>(deepinsight_module,
-                                    "InsightFaceRecognitionModel")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::deepinsight::InsightFaceRecognitionModel& self,
-              pybind11::array& data) {
-             auto mat = PyArrayToCvMat(data);
-             vision::FaceRecognitionResult res;
-             self.Predict(&mat, &res);
-             return res;
-           })
-      .def_readwrite("size",
-                     &vision::deepinsight::InsightFaceRecognitionModel::size)
-      .def_readwrite("alpha",
-                     &vision::deepinsight::InsightFaceRecognitionModel::alpha)
-      .def_readwrite("beta",
-                     &vision::deepinsight::InsightFaceRecognitionModel::beta)
-      .def_readwrite("swap_rb",
-                     &vision::deepinsight::InsightFaceRecognitionModel::swap_rb)
-      .def_readwrite(
-          "l2_normalize",
-          &vision::deepinsight::InsightFaceRecognitionModel::l2_normalize);
-  // Bind ArcFace
-  pybind11::class_<vision::deepinsight::ArcFace,
-                   vision::deepinsight::InsightFaceRecognitionModel>(
-      deepinsight_module, "ArcFace")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::deepinsight::ArcFace& self, pybind11::array& data) {
-             auto mat = PyArrayToCvMat(data);
-             vision::FaceRecognitionResult res;
-             self.Predict(&mat, &res);
-             return res;
-           })
-      .def_readwrite("size", &vision::deepinsight::ArcFace::size)
-      .def_readwrite("alpha", &vision::deepinsight::ArcFace::alpha)
-      .def_readwrite("beta", &vision::deepinsight::ArcFace::beta)
-      .def_readwrite("swap_rb", &vision::deepinsight::ArcFace::swap_rb)
-      .def_readwrite("l2_normalize",
-                     &vision::deepinsight::ArcFace::l2_normalize);
-  // Bind CosFace
-  pybind11::class_<vision::deepinsight::CosFace,
-                   vision::deepinsight::InsightFaceRecognitionModel>(
-      deepinsight_module, "CosFace")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::deepinsight::CosFace& self, pybind11::array& data) {
-             auto mat = PyArrayToCvMat(data);
-             vision::FaceRecognitionResult res;
-             self.Predict(&mat, &res);
-             return res;
-           })
-      .def_readwrite("size", &vision::deepinsight::CosFace::size)
-      .def_readwrite("alpha", &vision::deepinsight::CosFace::alpha)
-      .def_readwrite("beta", &vision::deepinsight::CosFace::beta)
-      .def_readwrite("swap_rb", &vision::deepinsight::CosFace::swap_rb)
-      .def_readwrite("l2_normalize",
-                     &vision::deepinsight::CosFace::l2_normalize);
-  // Bind Partial FC
-  pybind11::class_<vision::deepinsight::PartialFC,
-                   vision::deepinsight::InsightFaceRecognitionModel>(
-      deepinsight_module, "PartialFC")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::deepinsight::PartialFC& self, pybind11::array& data) {
-             auto mat = PyArrayToCvMat(data);
-             vision::FaceRecognitionResult res;
-             self.Predict(&mat, &res);
-             return res;
-           })
-      .def_readwrite("size", &vision::deepinsight::PartialFC::size)
-      .def_readwrite("alpha", &vision::deepinsight::PartialFC::alpha)
-      .def_readwrite("beta", &vision::deepinsight::PartialFC::beta)
-      .def_readwrite("swap_rb", &vision::deepinsight::PartialFC::swap_rb)
-      .def_readwrite("l2_normalize",
-                     &vision::deepinsight::PartialFC::l2_normalize);
-  // Bind VPL
-  pybind11::class_<vision::deepinsight::VPL,
-                   vision::deepinsight::InsightFaceRecognitionModel>(
-      deepinsight_module, "VPL")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::deepinsight::VPL& self, pybind11::array& data) {
-             auto mat = PyArrayToCvMat(data);
-             vision::FaceRecognitionResult res;
-             self.Predict(&mat, &res);
-             return res;
-           })
-      .def_readwrite("size", &vision::deepinsight::VPL::size)
-      .def_readwrite("alpha", &vision::deepinsight::VPL::alpha)
-      .def_readwrite("beta", &vision::deepinsight::VPL::beta)
-      .def_readwrite("swap_rb", &vision::deepinsight::VPL::swap_rb)
-      .def_readwrite("l2_normalize", &vision::deepinsight::VPL::l2_normalize);
-}
-
-}  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/nanodet_plus.cc

@@ -10,7 +10,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/rangilyu/nanodet_plus.h"
+#include "fastdeploy/vision/detection/contrib/nanodet_plus.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -18,7 +18,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace rangilyu {
+namespace detection {
 
 struct NanoDetPlusCenterPoint {
   int grid0;
@@ -350,6 +350,6 @@ bool NanoDetPlus::Predict(cv::Mat* im, DetectionResult* result,
   return true;
 }
 
-}  // namespace rangilyu
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/nanodet_plus.h

@@ -22,7 +22,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace rangilyu {
+namespace detection {
 
 class FASTDEPLOY_DECL NanoDetPlus : public FastDeployModel {
  public:
@@ -34,7 +34,7 @@ class FASTDEPLOY_DECL NanoDetPlus : public FastDeployModel {
               const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "RangiLyu/nanodet"; }
+  std::string ModelName() const { return "nanodet"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -96,6 +96,6 @@ class FASTDEPLOY_DECL NanoDetPlus : public FastDeployModel {
   bool is_dynamic_input_;
 };
 
-}  // namespace rangilyu
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/nanodet_plus_pybind.cc

@@ -15,27 +15,25 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindMegvii(pybind11::module& m) {
+void BindNanoDetPlus(pybind11::module& m) {
-  auto megvii_module =
+  pybind11::class_<vision::detection::NanoDetPlus, FastDeployModel>(
-      m.def_submodule("megvii", "https://github.com/megvii/YOLOX");
+      m, "NanoDetPlus")
-  pybind11::class_<vision::megvii::YOLOX, FastDeployModel>(
-      megvii_module, "YOLOX")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::megvii::YOLOX& self, pybind11::array& data,
+           [](vision::detection::NanoDetPlus& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::DetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::megvii::YOLOX::size)
+      .def_readwrite("size", &vision::detection::NanoDetPlus::size)
       .def_readwrite("padding_value",
-                     &vision::megvii::YOLOX::padding_value)
+                     &vision::detection::NanoDetPlus::padding_value)
-      .def_readwrite("is_decode_exported", 
+      .def_readwrite("keep_ratio", &vision::detection::NanoDetPlus::keep_ratio)
-                     &vision::megvii::YOLOX::is_decode_exported)
       .def_readwrite("downsample_strides",
-                     &vision::megvii::YOLOX::downsample_strides)
+                     &vision::detection::NanoDetPlus::downsample_strides)
-      .def_readwrite("max_wh", &vision::megvii::YOLOX::max_wh);
+      .def_readwrite("max_wh", &vision::detection::NanoDetPlus::max_wh)
+      .def_readwrite("reg_max", &vision::detection::NanoDetPlus::reg_max);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov5.cc

@@ -12,17 +12,17 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/ultralytics/yolov5.h"
+#include "fastdeploy/vision/detection/contrib/yolov5.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
 namespace fastdeploy {
 namespace vision {
-namespace ultralytics {
+namespace detection {
 
-void LetterBox(Mat* mat, std::vector<int> size, std::vector<float> color,
+void YOLOv5::LetterBox(Mat* mat, std::vector<int> size,
-               bool _auto, bool scale_fill = false, bool scale_up = true,
+                       std::vector<float> color, bool _auto, bool scale_fill,
-               int stride = 32) {
+                       bool scale_up, int stride) {
   float scale =
       std::min(size[1] * 1.0 / mat->Height(), size[0] * 1.0 / mat->Width());
   if (!scale_up) {
@@ -290,6 +290,6 @@ bool YOLOv5::Predict(cv::Mat* im, DetectionResult* result, float conf_threshold,
   return true;
 }
 
-}  // namespace ultralytics
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov5.h

@@ -19,7 +19,7 @@
 
 namespace fastdeploy {
 namespace vision {
-namespace ultralytics {
+namespace detection {
 
 class FASTDEPLOY_DECL YOLOv5 : public FastDeployModel {
  public:
@@ -30,7 +30,7 @@ class FASTDEPLOY_DECL YOLOv5 : public FastDeployModel {
          const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "ultralytics/yolov5"; }
+  std::string ModelName() const { return "yolov5"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -81,21 +81,28 @@ class FASTDEPLOY_DECL YOLOv5 : public FastDeployModel {
   // conf_threshold 后处理时过滤box的置信度阈值
   // nms_iou_threshold 后处理时NMS设定的iou阈值
   // multi_label 后处理时box选取是否采用多标签方式
-  bool Postprocess(
+  bool Postprocess(FDTensor& infer_result, DetectionResult* result,
-      FDTensor& infer_result, DetectionResult* result,
                    const std::map<std::string, std::array<float, 2>>& im_info,
-      float conf_threshold, float nms_iou_threshold, bool multi_label);
+                   float conf_threshold, float nms_iou_threshold,
+                   bool multi_label);
 
   // 查看输入是否为动态维度的 不建议直接使用 不同模型的逻辑可能不一致
   bool IsDynamicInput() const { return is_dynamic_input_; }
 
-  // whether to inference with dynamic shape (e.g ONNX export with dynamic shape or not.)
+  void LetterBox(Mat* mat, std::vector<int> size, std::vector<float> color,
-  // YOLOv5 official 'export_onnx.py' script will export dynamic ONNX by default.
+                 bool _auto, bool scale_fill = false, bool scale_up = true,
-  // while is_dynamic_shape if 'false', is_mini_pad will force 'false'. This value will
+                 int stride = 32);
+
+  // whether to inference with dynamic shape (e.g ONNX export with dynamic shape
+  // or not.)
+  // YOLOv5 official 'export_onnx.py' script will export dynamic ONNX by
+  // default.
+  // while is_dynamic_shape if 'false', is_mini_pad will force 'false'. This
+  // value will
   // auto check by fastdeploy after the internal Runtime already initialized.
   bool is_dynamic_input_;
 };
 
-}  // namespace ultralytics
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov5_pybind.cc

@@ -15,27 +15,24 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindMeituan(pybind11::module& m) {
+void BindYOLOv5(pybind11::module& m) {
-  auto meituan_module =
+  pybind11::class_<vision::detection::YOLOv5, FastDeployModel>(m, "YOLOv5")
-      m.def_submodule("meituan", "https://github.com/meituan/YOLOv6");
-  pybind11::class_<vision::meituan::YOLOv6, FastDeployModel>(
-      meituan_module, "YOLOv6")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::meituan::YOLOv6& self, pybind11::array& data,
+           [](vision::detection::YOLOv5& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::DetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::meituan::YOLOv6::size)
+      .def_readwrite("size", &vision::detection::YOLOv5::size)
-      .def_readwrite("padding_value",
+      .def_readwrite("padding_value", &vision::detection::YOLOv5::padding_value)
-                     &vision::meituan::YOLOv6::padding_value)
+      .def_readwrite("is_mini_pad", &vision::detection::YOLOv5::is_mini_pad)
-      .def_readwrite("is_mini_pad", &vision::meituan::YOLOv6::is_mini_pad)
+      .def_readwrite("is_no_pad", &vision::detection::YOLOv5::is_no_pad)
-      .def_readwrite("is_no_pad", &vision::meituan::YOLOv6::is_no_pad)
+      .def_readwrite("is_scale_up", &vision::detection::YOLOv5::is_scale_up)
-      .def_readwrite("is_scale_up", &vision::meituan::YOLOv6::is_scale_up)
+      .def_readwrite("stride", &vision::detection::YOLOv5::stride)
-      .def_readwrite("stride", &vision::meituan::YOLOv6::stride)
+      .def_readwrite("max_wh", &vision::detection::YOLOv5::max_wh)
-      .def_readwrite("max_wh", &vision::meituan::YOLOv6::max_wh);
+      .def_readwrite("multi_label", &vision::detection::YOLOv5::multi_label);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov5lite.cc

@@ -12,13 +12,13 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/ppogg/yolov5lite.h"
+#include "fastdeploy/vision/detection/contrib/yolov5lite.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
 namespace fastdeploy {
 namespace vision {
-namespace ppogg {
+namespace detection {
 
 void YOLOv5Lite::LetterBox(Mat* mat, const std::vector<int>& size,
                            const std::vector<float>& color, bool _auto,
@@ -191,7 +191,6 @@ bool YOLOv5Lite::PostprocessWithDecode(
   std::vector<YOLOv5Lite::Anchor> anchors;
   int num_anchors = anchor_config[0].size() / 2;
   GenerateAnchors(size, downsample_strides, &anchors, num_anchors);
-
   // infer_result shape might look like (1,n,85=5+80)
   float* data = static_cast<float*>(infer_result.Data());
   for (size_t i = 0; i < infer_result.shape[1]; ++i) {
@@ -395,6 +394,6 @@ bool YOLOv5Lite::Predict(cv::Mat* im, DetectionResult* result,
   return true;
 }
 
-}  // namespace ppogg
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov5lite.h

@@ -19,7 +19,7 @@
 
 namespace fastdeploy {
 namespace vision {
-namespace ppogg {
+namespace detection {
 
 class FASTDEPLOY_DECL YOLOv5Lite : public FastDeployModel {
  public:
@@ -30,8 +30,7 @@ class FASTDEPLOY_DECL YOLOv5Lite : public FastDeployModel {
              const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  virtual std::string ModelName() const { return "ppogg/YOLOv5-Lite"; }
+  virtual std::string ModelName() const { return "YOLOv5-Lite"; }
-
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
   // result 为模型预测的输出结构体
@@ -134,6 +133,6 @@ class FASTDEPLOY_DECL YOLOv5Lite : public FastDeployModel {
   // auto check by fastdeploy after the internal Runtime already initialized.
   bool is_dynamic_input_;
 };
-}  // namespace ppogg
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov5lite_pybind.cc

@@ -15,29 +15,29 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindPpogg(pybind11::module& m) {
+void BindYOLOv5Lite(pybind11::module& m) {
-  auto ppogg_module =
+  pybind11::class_<vision::detection::YOLOv5Lite, FastDeployModel>(m,
-      m.def_submodule("ppogg", "https://github.com/ppogg/YOLOv5-Lite");
-  pybind11::class_<vision::ppogg::YOLOv5Lite, FastDeployModel>(ppogg_module,
                                                                    "YOLOv5Lite")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::ppogg::YOLOv5Lite& self, pybind11::array& data,
+           [](vision::detection::YOLOv5Lite& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::DetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::ppogg::YOLOv5Lite::size)
+      .def_readwrite("size", &vision::detection::YOLOv5Lite::size)
-      .def_readwrite("padding_value", &vision::ppogg::YOLOv5Lite::padding_value)
+      .def_readwrite("padding_value",
-      .def_readwrite("is_mini_pad", &vision::ppogg::YOLOv5Lite::is_mini_pad)
+                     &vision::detection::YOLOv5Lite::padding_value)
-      .def_readwrite("is_no_pad", &vision::ppogg::YOLOv5Lite::is_no_pad)
+      .def_readwrite("is_mini_pad", &vision::detection::YOLOv5Lite::is_mini_pad)
-      .def_readwrite("is_scale_up", &vision::ppogg::YOLOv5Lite::is_scale_up)
+      .def_readwrite("is_no_pad", &vision::detection::YOLOv5Lite::is_no_pad)
-      .def_readwrite("stride", &vision::ppogg::YOLOv5Lite::stride)
+      .def_readwrite("is_scale_up", &vision::detection::YOLOv5Lite::is_scale_up)
-      .def_readwrite("max_wh", &vision::ppogg::YOLOv5Lite::max_wh)
+      .def_readwrite("stride", &vision::detection::YOLOv5Lite::stride)
-      .def_readwrite("anchor_config", &vision::ppogg::YOLOv5Lite::anchor_config)
+      .def_readwrite("max_wh", &vision::detection::YOLOv5Lite::max_wh)
+      .def_readwrite("anchor_config",
+                     &vision::detection::YOLOv5Lite::anchor_config)
       .def_readwrite("is_decode_exported",
-                     &vision::ppogg::YOLOv5Lite::is_decode_exported);
+                     &vision::detection::YOLOv5Lite::is_decode_exported);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov6.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/meituan/yolov6.h"
+#include "fastdeploy/vision/detection/contrib/yolov6.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,11 +20,11 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace meituan {
+namespace detection {
 
-void LetterBox(Mat* mat, std::vector<int> size, std::vector<float> color,
+void YOLOv6::LetterBox(Mat* mat, std::vector<int> size,
-               bool _auto, bool scale_fill = false, bool scale_up = true,
+                       std::vector<float> color, bool _auto, bool scale_fill,
-               int stride = 32) {
+                       bool scale_up, int stride) {
   float scale = std::min(size[1] * 1.0f / static_cast<float>(mat->Height()),
                          size[0] * 1.0f / static_cast<float>(mat->Width()));
   if (!scale_up) {
@@ -262,6 +262,6 @@ bool YOLOv6::Predict(cv::Mat* im, DetectionResult* result, float conf_threshold,
   return true;
 }
 
-}  // namespace meituan
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov6.h

@@ -22,7 +22,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace meituan {
+namespace detection {
 
 class FASTDEPLOY_DECL YOLOv6 : public FastDeployModel {
  public:
@@ -33,7 +33,7 @@ class FASTDEPLOY_DECL YOLOv6 : public FastDeployModel {
          const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "meituan/YOLOv6"; }
+  std::string ModelName() const { return "YOLOv6"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -61,7 +61,8 @@ class FASTDEPLOY_DECL YOLOv6 : public FastDeployModel {
   bool is_scale_up;
   // padding stride, for is_mini_pad
   int stride;
-  // for offseting the boxes by classes when using NMS, default 4096 in meituan/YOLOv6
+  // for offseting the boxes by classes when using NMS, default 4096 in
+  // meituan/YOLOv6
   float max_wh;
 
  private:
@@ -81,21 +82,27 @@ class FASTDEPLOY_DECL YOLOv6 : public FastDeployModel {
   // im_info 为预处理记录的信息，后处理用于还原box
   // conf_threshold 后处理时过滤box的置信度阈值
   // nms_iou_threshold 后处理时NMS设定的iou阈值
-  bool Postprocess(
+  bool Postprocess(FDTensor& infer_result, DetectionResult* result,
-      FDTensor& infer_result, DetectionResult* result,
                    const std::map<std::string, std::array<float, 2>>& im_info,
                    float conf_threshold, float nms_iou_threshold);
 
   // 查看输入是否为动态维度的 不建议直接使用 不同模型的逻辑可能不一致
   bool IsDynamicInput() const { return is_dynamic_input_; }
 
-  // whether to inference with dynamic shape (e.g ONNX export with dynamic shape or not.)
+  void LetterBox(Mat* mat, std::vector<int> size, std::vector<float> color,
-  // meituan/YOLOv6 official 'export_onnx.py' script will export static ONNX by default.
+                 bool _auto, bool scale_fill = false, bool scale_up = true,
-  // while is_dynamic_input if 'false', is_mini_pad will force 'false'. This value will
+                 int stride = 32);
+
+  // whether to inference with dynamic shape (e.g ONNX export with dynamic shape
+  // or not.)
+  // meituan/YOLOv6 official 'export_onnx.py' script will export static ONNX by
+  // default.
+  // while is_dynamic_input if 'false', is_mini_pad will force 'false'. This
+  // value will
   // auto check by fastdeploy after the internal Runtime already initialized.
   bool is_dynamic_input_;
 };
 
-}  // namespace meituan
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolov6_pybind.cc

@@ -0,0 +1,37 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindYOLOv6(pybind11::module& m) {
+  pybind11::class_<vision::detection::YOLOv6, FastDeployModel>(m, "YOLOv6")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::detection::YOLOv6& self, pybind11::array& data,
+              float conf_threshold, float nms_iou_threshold) {
+             auto mat = PyArrayToCvMat(data);
+             vision::DetectionResult res;
+             self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
+             return res;
+           })
+      .def_readwrite("size", &vision::detection::YOLOv6::size)
+      .def_readwrite("padding_value", &vision::detection::YOLOv6::padding_value)
+      .def_readwrite("is_mini_pad", &vision::detection::YOLOv6::is_mini_pad)
+      .def_readwrite("is_no_pad", &vision::detection::YOLOv6::is_no_pad)
+      .def_readwrite("is_scale_up", &vision::detection::YOLOv6::is_scale_up)
+      .def_readwrite("stride", &vision::detection::YOLOv6::stride)
+      .def_readwrite("max_wh", &vision::detection::YOLOv6::max_wh);
+}
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolox.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/megvii/yolox.h"
+#include "fastdeploy/vision/detection/contrib/yolox.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace megvii {
+namespace detection {
 
 struct YOLOXAnchor {
   int grid0;
@@ -334,6 +334,6 @@ bool YOLOX::Predict(cv::Mat* im, DetectionResult* result, float conf_threshold,
   return true;
 }
 
-}  // namespace megvii
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolox.h

@@ -22,7 +22,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace megvii {
+namespace detection {
 
 class FASTDEPLOY_DECL YOLOX : public FastDeployModel {
  public:
@@ -33,7 +33,7 @@ class FASTDEPLOY_DECL YOLOX : public FastDeployModel {
         const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "megvii/YOLOX"; }
+  std::string ModelName() const { return "YOLOX"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -79,8 +79,7 @@ class FASTDEPLOY_DECL YOLOX : public FastDeployModel {
   // im_info 为预处理记录的信息，后处理用于还原box
   // conf_threshold 后处理时过滤box的置信度阈值
   // nms_iou_threshold 后处理时NMS设定的iou阈值
-  bool Postprocess(
+  bool Postprocess(FDTensor& infer_result, DetectionResult* result,
-      FDTensor& infer_result, DetectionResult* result,
                    const std::map<std::string, std::array<float, 2>>& im_info,
                    float conf_threshold, float nms_iou_threshold);
 
@@ -93,13 +92,16 @@ class FASTDEPLOY_DECL YOLOX : public FastDeployModel {
   // 查看输入是否为动态维度的 不建议直接使用 不同模型的逻辑可能不一致
   bool IsDynamicInput() const { return is_dynamic_input_; }
 
-  // whether to inference with dynamic shape (e.g ONNX export with dynamic shape or not.)
+  // whether to inference with dynamic shape (e.g ONNX export with dynamic shape
-  // megvii/YOLOX official 'export_onnx.py' script will export static ONNX by default.
+  // or not.)
-  // while is_dynamic_shape if 'false', is_mini_pad will force 'false'. This value will
+  // megvii/YOLOX official 'export_onnx.py' script will export static ONNX by
+  // default.
+  // while is_dynamic_shape if 'false', is_mini_pad will force 'false'. This
+  // value will
   // auto check by fastdeploy after the internal Runtime already initialized.
   bool is_dynamic_input_;
 };
 
-}  // namespace megvii
+}  // namespace detection
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/contrib/yolox_pybind.cc

@@ -15,27 +15,23 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindRangiLyu(pybind11::module& m) {
+void BindYOLOX(pybind11::module& m) {
-  auto rangilyu_module =
+  pybind11::class_<vision::detection::YOLOX, FastDeployModel>(m, "YOLOX")
-      m.def_submodule("rangilyu", "https://github.com/RangiLyu/nanodet");
-  pybind11::class_<vision::rangilyu::NanoDetPlus, FastDeployModel>(
-      rangilyu_module, "NanoDetPlus")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::rangilyu::NanoDetPlus& self, pybind11::array& data,
+           [](vision::detection::YOLOX& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::DetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::rangilyu::NanoDetPlus::size)
+      .def_readwrite("size", &vision::detection::YOLOX::size)
-      .def_readwrite("padding_value",
+      .def_readwrite("padding_value", &vision::detection::YOLOX::padding_value)
-                     &vision::rangilyu::NanoDetPlus::padding_value)
+      .def_readwrite("is_decode_exported",
-      .def_readwrite("keep_ratio", &vision::rangilyu::NanoDetPlus::keep_ratio)
+                     &vision::detection::YOLOX::is_decode_exported)
       .def_readwrite("downsample_strides",
-                     &vision::rangilyu::NanoDetPlus::downsample_strides)
+                     &vision::detection::YOLOX::downsample_strides)
-      .def_readwrite("max_wh", &vision::rangilyu::NanoDetPlus::max_wh)
+      .def_readwrite("max_wh", &vision::detection::YOLOX::max_wh);
-      .def_readwrite("reg_max", &vision::rangilyu::NanoDetPlus::reg_max);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/detection/detection_pybind.cc

@@ -19,6 +19,11 @@ namespace fastdeploy {
 void BindYOLOv7(pybind11::module& m);
 void BindScaledYOLOv4(pybind11::module& m);
 void BindYOLOR(pybind11::module& m);
+void BindYOLOv6(pybind11::module& m);
+void BindYOLOv5Lite(pybind11::module& m);
+void BindYOLOv5(pybind11::module& m);
+void BindYOLOX(pybind11::module& m);
+void BindNanoDetPlus(pybind11::module& m);
 void BindPPDet(pybind11::module& m);
 
 void BindDetection(pybind11::module& m) {
@@ -28,5 +33,10 @@ void BindDetection(pybind11::module& m) {
   BindYOLOv7(detection_module);
   BindScaledYOLOv4(detection_module);
   BindYOLOR(detection_module);
+  BindYOLOv6(detection_module);
+  BindYOLOv5Lite(detection_module);
+  BindYOLOv5(detection_module);
+  BindYOLOX(detection_module);
+  BindNanoDetPlus(detection_module);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/retinaface.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/biubug6/retinaface.h"
+#include "fastdeploy/vision/facedet/contrib/retinaface.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace biubug6 {
+namespace facedet {
 
 struct RetinaAnchor {
   float cx;
@@ -305,6 +305,6 @@ bool RetinaFace::Predict(cv::Mat* im, FaceDetectionResult* result,
   return true;
 }
 
-}  // namespace biubug6
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/retinaface.h

@@ -21,7 +21,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace biubug6 {
+namespace facedet {
 
 class FASTDEPLOY_DECL RetinaFace : public FastDeployModel {
  public:
@@ -32,7 +32,7 @@ class FASTDEPLOY_DECL RetinaFace : public FastDeployModel {
              const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "biubug6/Pytorch_Retinaface"; }
+  std::string ModelName() const { return "Pytorch_Retinaface"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -87,6 +87,6 @@ class FASTDEPLOY_DECL RetinaFace : public FastDeployModel {
   bool is_dynamic_input_;
 };
 
-}  // namespace biubug6
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/retinaface_pybind.cc

@@ -15,26 +15,24 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindBiubug6(pybind11::module& m) {
+void BindRetinaFace(pybind11::module& m) {
-  auto biubug6_module = m.def_submodule(
+  pybind11::class_<vision::facedet::RetinaFace, FastDeployModel>(m,
-      "biubug6", "https://github.com/biubug6/Pytorch_Retinaface");
-  pybind11::class_<vision::biubug6::RetinaFace, FastDeployModel>(biubug6_module,
                                                                  "RetinaFace")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::biubug6::RetinaFace& self, pybind11::array& data,
+           [](vision::facedet::RetinaFace& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::FaceDetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::biubug6::RetinaFace::size)
+      .def_readwrite("size", &vision::facedet::RetinaFace::size)
-      .def_readwrite("variance", &vision::biubug6::RetinaFace::variance)
+      .def_readwrite("variance", &vision::facedet::RetinaFace::variance)
       .def_readwrite("downsample_strides",
-                     &vision::biubug6::RetinaFace::downsample_strides)
+                     &vision::facedet::RetinaFace::downsample_strides)
-      .def_readwrite("min_sizes", &vision::biubug6::RetinaFace::min_sizes)
+      .def_readwrite("min_sizes", &vision::facedet::RetinaFace::min_sizes)
       .def_readwrite("landmarks_per_face",
-                     &vision::biubug6::RetinaFace::landmarks_per_face);
+                     &vision::facedet::RetinaFace::landmarks_per_face);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/scrfd.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepinsight/scrfd.h"
+#include "fastdeploy/vision/facedet/contrib/scrfd.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace facedet {
 
 void SCRFD::LetterBox(Mat* mat, const std::vector<int>& size,
                       const std::vector<float>& color, bool _auto,
@@ -246,14 +246,18 @@ bool SCRFD::Postprocess(
       float r = offsets[2];  // right
       float b = offsets[3];  // bottom
 
-      float x1 =
+      float x1 = ((cx - l) * static_cast<float>(current_stride) -
-          ((cx - l) * static_cast<float>(current_stride) - static_cast<float>(pad_w)) / scale;  // cx - l x1
+                  static_cast<float>(pad_w)) /
-      float y1 =
+                 scale;  // cx - l x1
-          ((cy - t) * static_cast<float>(current_stride) - static_cast<float>(pad_h)) / scale;  // cy - t y1
+      float y1 = ((cy - t) * static_cast<float>(current_stride) -
-      float x2 =
+                  static_cast<float>(pad_h)) /
-          ((cx + r) * static_cast<float>(current_stride) - static_cast<float>(pad_w)) / scale;  // cx + r x2
+                 scale;  // cy - t y1
-      float y2 =
+      float x2 = ((cx + r) * static_cast<float>(current_stride) -
-          ((cy + b) * static_cast<float>(current_stride) - static_cast<float>(pad_h)) / scale;  // cy + b y2
+                  static_cast<float>(pad_w)) /
+                 scale;  // cx + r x2
+      float y2 = ((cy + b) * static_cast<float>(current_stride) -
+                  static_cast<float>(pad_h)) /
+                 scale;  // cy + b y2
       result->boxes.emplace_back(std::array<float, 4>{x1, y1, x2, y2});
       result->scores.push_back(cls_conf);
       if (use_kps) {
@@ -264,9 +268,11 @@ bool SCRFD::Postprocess(
         for (unsigned int j = 0; j < landmarks_per_face * 2; j += 2) {
           float kps_l = kps_offsets[j];
           float kps_t = kps_offsets[j + 1];
-          float kps_x = ((cx + kps_l) * static_cast<float>(current_stride) - static_cast<float>(pad_w)) /
+          float kps_x = ((cx + kps_l) * static_cast<float>(current_stride) -
+                         static_cast<float>(pad_w)) /
                         scale;  // cx + l x
-          float kps_y = ((cy + kps_t) * static_cast<float>(current_stride) - static_cast<float>(pad_h)) /
+          float kps_y = ((cy + kps_t) * static_cast<float>(current_stride) -
+                         static_cast<float>(pad_h)) /
                         scale;  // cy + t y
           result->landmarks.emplace_back(std::array<float, 2>{kps_x, kps_y});
         }
@@ -358,6 +364,6 @@ bool SCRFD::Predict(cv::Mat* im, FaceDetectionResult* result,
   return true;
 }
 
-}  // namespace deepinsight
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/scrfd.h

@@ -22,7 +22,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace facedet {
 
 class FASTDEPLOY_DECL SCRFD : public FastDeployModel {
  public:
@@ -33,7 +33,7 @@ class FASTDEPLOY_DECL SCRFD : public FastDeployModel {
         const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "deepinsight/scrfd"; }
+  std::string ModelName() const { return "scrfd"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -117,6 +117,6 @@ class FASTDEPLOY_DECL SCRFD : public FastDeployModel {
 
   std::unordered_map<int, std::vector<SCRFDPoint>> center_points_;
 };
-}  // namespace deepinsight
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/scrfd_pybind.cc

@@ -0,0 +1,45 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindSCRFD(pybind11::module& m) {
+  // Bind SCRFD
+  pybind11::class_<vision::facedet::SCRFD, FastDeployModel>(m, "SCRFD")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::facedet::SCRFD& self, pybind11::array& data,
+              float conf_threshold, float nms_iou_threshold) {
+             auto mat = PyArrayToCvMat(data);
+             vision::FaceDetectionResult res;
+             self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
+             return res;
+           })
+      .def_readwrite("size", &vision::facedet::SCRFD::size)
+      .def_readwrite("padding_value", &vision::facedet::SCRFD::padding_value)
+      .def_readwrite("is_mini_pad", &vision::facedet::SCRFD::is_mini_pad)
+      .def_readwrite("is_no_pad", &vision::facedet::SCRFD::is_no_pad)
+      .def_readwrite("is_scale_up", &vision::facedet::SCRFD::is_scale_up)
+      .def_readwrite("stride", &vision::facedet::SCRFD::stride)
+      .def_readwrite("use_kps", &vision::facedet::SCRFD::use_kps)
+      .def_readwrite("max_nms", &vision::facedet::SCRFD::max_nms)
+      .def_readwrite("downsample_strides",
+                     &vision::facedet::SCRFD::downsample_strides)
+      .def_readwrite("num_anchors", &vision::facedet::SCRFD::num_anchors)
+      .def_readwrite("landmarks_per_face",
+                     &vision::facedet::SCRFD::landmarks_per_face);
+}
+
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/ultraface.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/linzaer/ultraface.h"
+#include "fastdeploy/vision/facedet/contrib/ultraface.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace linzaer {
+namespace facedet {
 
 UltraFace::UltraFace(const std::string& model_file,
                      const std::string& params_file,
@@ -216,6 +216,6 @@ bool UltraFace::Predict(cv::Mat* im, FaceDetectionResult* result,
   return true;
 }
 
-}  // namespace linzaer
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/ultraface.h

@@ -21,7 +21,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace linzaer {
+namespace facedet {
 
 class FASTDEPLOY_DECL UltraFace : public FastDeployModel {
  public:
@@ -79,6 +79,6 @@ class FASTDEPLOY_DECL UltraFace : public FastDeployModel {
   bool is_dynamic_input_;
 };
 
-}  // namespace linzaer
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/ultraface_pybind.cc

@@ -15,21 +15,17 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindLinzaer(pybind11::module& m) {
+void BindUltraFace(pybind11::module& m) {
-  auto linzaer_module = m.def_submodule(
+  pybind11::class_<vision::facedet::UltraFace, FastDeployModel>(m, "UltraFace")
-      "linzaer",
-      "https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB");
-  pybind11::class_<vision::linzaer::UltraFace, FastDeployModel>(linzaer_module,
-                                                                "UltraFace")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::linzaer::UltraFace& self, pybind11::array& data,
+           [](vision::facedet::UltraFace& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::FaceDetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::linzaer::UltraFace::size);
+      .def_readwrite("size", &vision::facedet::UltraFace::size);
 }
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/yolov5face.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepcam/yolov5face.h"
+#include "fastdeploy/vision/facedet/contrib/yolov5face.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepcam {
+namespace facedet {
 
 void LetterBox(Mat* mat, std::vector<int> size, std::vector<float> color,
                bool _auto, bool scale_fill = false, bool scale_up = true,
@@ -289,6 +289,6 @@ bool YOLOv5Face::Predict(cv::Mat* im, FaceDetectionResult* result,
   return true;
 }
 
-}  // namespace deepcam
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/yolov5face.h

@@ -21,7 +21,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepcam {
+namespace facedet {
 
 class FASTDEPLOY_DECL YOLOv5Face : public FastDeployModel {
  public:
@@ -32,7 +32,7 @@ class FASTDEPLOY_DECL YOLOv5Face : public FastDeployModel {
              const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "deepcam-cn/yolov5-face"; }
+  std::string ModelName() const { return "yolov5-face"; }
 
   // 模型预测接口，即用户调用的接口
   // im 为用户的输入数据，目前对于CV均定义为cv::Mat
@@ -92,6 +92,6 @@ class FASTDEPLOY_DECL YOLOv5Face : public FastDeployModel {
   bool is_dynamic_input_;
 };
 
-}  // namespace deepcam
+}  // namespace facedet
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/contrib/yolov5face_pybind.cc

@@ -15,29 +15,27 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindDeepCam(pybind11::module& m) {
+void BindYOLOv5Face(pybind11::module& m) {
-  auto deepcam_module =
+  pybind11::class_<vision::facedet::YOLOv5Face, FastDeployModel>(m,
-      m.def_submodule("deepcam", "https://github.com/deepcam-cn/yolov5-face");
-  pybind11::class_<vision::deepcam::YOLOv5Face, FastDeployModel>(deepcam_module,
                                                                  "YOLOv5Face")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::deepcam::YOLOv5Face& self, pybind11::array& data,
+           [](vision::facedet::YOLOv5Face& self, pybind11::array& data,
               float conf_threshold, float nms_iou_threshold) {
              auto mat = PyArrayToCvMat(data);
              vision::FaceDetectionResult res;
              self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
              return res;
            })
-      .def_readwrite("size", &vision::deepcam::YOLOv5Face::size)
+      .def_readwrite("size", &vision::facedet::YOLOv5Face::size)
       .def_readwrite("padding_value",
-                     &vision::deepcam::YOLOv5Face::padding_value)
+                     &vision::facedet::YOLOv5Face::padding_value)
-      .def_readwrite("is_mini_pad", &vision::deepcam::YOLOv5Face::is_mini_pad)
+      .def_readwrite("is_mini_pad", &vision::facedet::YOLOv5Face::is_mini_pad)
-      .def_readwrite("is_no_pad", &vision::deepcam::YOLOv5Face::is_no_pad)
+      .def_readwrite("is_no_pad", &vision::facedet::YOLOv5Face::is_no_pad)
-      .def_readwrite("is_scale_up", &vision::deepcam::YOLOv5Face::is_scale_up)
+      .def_readwrite("is_scale_up", &vision::facedet::YOLOv5Face::is_scale_up)
-      .def_readwrite("stride", &vision::deepcam::YOLOv5Face::stride)
+      .def_readwrite("stride", &vision::facedet::YOLOv5Face::stride)
       .def_readwrite("landmarks_per_face",
-                     &vision::deepcam::YOLOv5Face::landmarks_per_face);
+                     &vision::facedet::YOLOv5Face::landmarks_per_face);
 }
 
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/facedet/facedet_pybind.cc

@@ -0,0 +1,31 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+
+void BindRetinaFace(pybind11::module& m);
+void BindUltraFace(pybind11::module& m);
+void BindYOLOv5Face(pybind11::module& m);
+void BindSCRFD(pybind11::module& m);
+
+void BindFaceDet(pybind11::module& m) {
+  auto facedet_module = m.def_submodule("facedet", "Face detection models.");
+  BindRetinaFace(facedet_module);
+  BindUltraFace(facedet_module);
+  BindYOLOv5Face(facedet_module);
+  BindSCRFD(facedet_module);
+}
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/arcface.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepinsight/arcface.h"
+#include "fastdeploy/vision/faceid/contrib/arcface.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 ArcFace::ArcFace(const std::string& model_file, const std::string& params_file,
                  const RuntimeOption& custom_option,
@@ -78,6 +78,6 @@ bool ArcFace::Predict(cv::Mat* im, FaceRecognitionResult* result) {
   return InsightFaceRecognitionModel::Predict(im, result);
 }
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/arcface.h

@@ -16,13 +16,13 @@
 #include "fastdeploy/fastdeploy_model.h"
 #include "fastdeploy/vision/common/processors/transform.h"
 #include "fastdeploy/vision/common/result.h"
-#include "fastdeploy/vision/deepinsight/insightface_rec.h"
+#include "fastdeploy/vision/faceid/contrib/insightface_rec.h"
 
 namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 class FASTDEPLOY_DECL ArcFace : public InsightFaceRecognitionModel {
  public:
@@ -60,6 +60,6 @@ class FASTDEPLOY_DECL ArcFace : public InsightFaceRecognitionModel {
                    FaceRecognitionResult* result) override;
 };
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/arcface_pybind.cc

@@ -0,0 +1,37 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindArcFace(pybind11::module& m) {
+  // Bind ArcFace
+  pybind11::class_<vision::faceid::ArcFace,
+                   vision::faceid::InsightFaceRecognitionModel>(m, "ArcFace")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::faceid::ArcFace& self, pybind11::array& data) {
+             auto mat = PyArrayToCvMat(data);
+             vision::FaceRecognitionResult res;
+             self.Predict(&mat, &res);
+             return res;
+           })
+      .def_readwrite("size", &vision::faceid::ArcFace::size)
+      .def_readwrite("alpha", &vision::faceid::ArcFace::alpha)
+      .def_readwrite("beta", &vision::faceid::ArcFace::beta)
+      .def_readwrite("swap_rb", &vision::faceid::ArcFace::swap_rb)
+      .def_readwrite("l2_normalize", &vision::faceid::ArcFace::l2_normalize);
+}
+
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/cosface.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepinsight/cosface.h"
+#include "fastdeploy/vision/faceid/contrib/cosface.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 CosFace::CosFace(const std::string& model_file, const std::string& params_file,
                  const RuntimeOption& custom_option,
@@ -78,6 +78,6 @@ bool CosFace::Predict(cv::Mat* im, FaceRecognitionResult* result) {
   return InsightFaceRecognitionModel::Predict(im, result);
 }
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/cosface.h

@@ -16,13 +16,13 @@
 #include "fastdeploy/fastdeploy_model.h"
 #include "fastdeploy/vision/common/processors/transform.h"
 #include "fastdeploy/vision/common/result.h"
-#include "fastdeploy/vision/deepinsight/insightface_rec.h"
+#include "fastdeploy/vision/faceid/contrib/insightface_rec.h"
 
 namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 class FASTDEPLOY_DECL CosFace : public InsightFaceRecognitionModel {
  public:
@@ -61,6 +61,6 @@ class FASTDEPLOY_DECL CosFace : public InsightFaceRecognitionModel {
                    FaceRecognitionResult* result) override;
 };
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/cosface_pybind.cc

@@ -0,0 +1,37 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindCosFace(pybind11::module& m) {
+  // Bind CosFace
+  pybind11::class_<vision::faceid::CosFace,
+                   vision::faceid::InsightFaceRecognitionModel>(m, "CosFace")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::faceid::CosFace& self, pybind11::array& data) {
+             auto mat = PyArrayToCvMat(data);
+             vision::FaceRecognitionResult res;
+             self.Predict(&mat, &res);
+             return res;
+           })
+      .def_readwrite("size", &vision::faceid::CosFace::size)
+      .def_readwrite("alpha", &vision::faceid::CosFace::alpha)
+      .def_readwrite("beta", &vision::faceid::CosFace::beta)
+      .def_readwrite("swap_rb", &vision::faceid::CosFace::swap_rb)
+      .def_readwrite("l2_normalize", &vision::faceid::CosFace::l2_normalize);
+}
+
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/insightface_rec.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepinsight/insightface_rec.h"
+#include "fastdeploy/vision/faceid/contrib/insightface_rec.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 InsightFaceRecognitionModel::InsightFaceRecognitionModel(
     const std::string& model_file, const std::string& params_file,
@@ -148,6 +148,6 @@ bool InsightFaceRecognitionModel::Predict(cv::Mat* im,
   return true;
 }
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/insightface_rec.h

@@ -21,7 +21,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 class FASTDEPLOY_DECL InsightFaceRecognitionModel : public FastDeployModel {
  public:
@@ -67,6 +67,6 @@ class FASTDEPLOY_DECL InsightFaceRecognitionModel : public FastDeployModel {
                            FaceRecognitionResult* result);
 };
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/insightface_rec_pybind.cc

@@ -0,0 +1,42 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindInsightFaceRecognitionModel(pybind11::module& m) {
+  // Bind InsightFaceRecognitionModel
+  pybind11::class_<vision::faceid::InsightFaceRecognitionModel,
+                   FastDeployModel>(m, "InsightFaceRecognitionModel")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::faceid::InsightFaceRecognitionModel& self,
+              pybind11::array& data) {
+             auto mat = PyArrayToCvMat(data);
+             vision::FaceRecognitionResult res;
+             self.Predict(&mat, &res);
+             return res;
+           })
+      .def_readwrite("size", &vision::faceid::InsightFaceRecognitionModel::size)
+      .def_readwrite("alpha",
+                     &vision::faceid::InsightFaceRecognitionModel::alpha)
+      .def_readwrite("beta", &vision::faceid::InsightFaceRecognitionModel::beta)
+      .def_readwrite("swap_rb",
+                     &vision::faceid::InsightFaceRecognitionModel::swap_rb)
+      .def_readwrite(
+          "l2_normalize",
+          &vision::faceid::InsightFaceRecognitionModel::l2_normalize);
+}
+
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/partial_fc.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepinsight/partial_fc.h"
+#include "fastdeploy/vision/faceid/contrib/partial_fc.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 PartialFC::PartialFC(const std::string& model_file,
                      const std::string& params_file,
@@ -79,6 +79,6 @@ bool PartialFC::Predict(cv::Mat* im, FaceRecognitionResult* result) {
   return InsightFaceRecognitionModel::Predict(im, result);
 }
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/partial_fc.h

@@ -16,13 +16,13 @@
 #include "fastdeploy/fastdeploy_model.h"
 #include "fastdeploy/vision/common/processors/transform.h"
 #include "fastdeploy/vision/common/result.h"
-#include "fastdeploy/vision/deepinsight/insightface_rec.h"
+#include "fastdeploy/vision/faceid/contrib/insightface_rec.h"
 
 namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 class FASTDEPLOY_DECL PartialFC : public InsightFaceRecognitionModel {
  public:
@@ -59,6 +59,6 @@ class FASTDEPLOY_DECL PartialFC : public InsightFaceRecognitionModel {
                    FaceRecognitionResult* result) override;
 };
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/partial_fc_pybind.cc

@@ -0,0 +1,37 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindPartialFC(pybind11::module& m) {
+  // Bind Partial FC
+  pybind11::class_<vision::faceid::PartialFC,
+                   vision::faceid::InsightFaceRecognitionModel>(m, "PartialFC")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::faceid::PartialFC& self, pybind11::array& data) {
+             auto mat = PyArrayToCvMat(data);
+             vision::FaceRecognitionResult res;
+             self.Predict(&mat, &res);
+             return res;
+           })
+      .def_readwrite("size", &vision::faceid::PartialFC::size)
+      .def_readwrite("alpha", &vision::faceid::PartialFC::alpha)
+      .def_readwrite("beta", &vision::faceid::PartialFC::beta)
+      .def_readwrite("swap_rb", &vision::faceid::PartialFC::swap_rb)
+      .def_readwrite("l2_normalize", &vision::faceid::PartialFC::l2_normalize);
+}
+
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/vpl.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/deepinsight/vpl.h"
+#include "fastdeploy/vision/faceid/contrib/vpl.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 VPL::VPL(const std::string& model_file, const std::string& params_file,
          const RuntimeOption& custom_option, const Frontend& model_format)
@@ -77,6 +77,6 @@ bool VPL::Predict(cv::Mat* im, FaceRecognitionResult* result) {
   return InsightFaceRecognitionModel::Predict(im, result);
 }
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/vpl.h

@@ -16,13 +16,13 @@
 #include "fastdeploy/fastdeploy_model.h"
 #include "fastdeploy/vision/common/processors/transform.h"
 #include "fastdeploy/vision/common/result.h"
-#include "fastdeploy/vision/deepinsight/insightface_rec.h"
+#include "fastdeploy/vision/faceid/contrib/insightface_rec.h"
 
 namespace fastdeploy {
 
 namespace vision {
 
-namespace deepinsight {
+namespace faceid {
 
 class FASTDEPLOY_DECL VPL : public InsightFaceRecognitionModel {
  public:
@@ -60,6 +60,6 @@ class FASTDEPLOY_DECL VPL : public InsightFaceRecognitionModel {
                    FaceRecognitionResult* result) override;
 };
 
-}  // namespace deepinsight
+}  // namespace faceid
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/contrib/vpl_pybind.cc

@@ -0,0 +1,37 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+void BindVPL(pybind11::module& m) {
+  // Bind VPL
+  pybind11::class_<vision::faceid::VPL,
+                   vision::faceid::InsightFaceRecognitionModel>(m, "VPL")
+      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
+      .def("predict",
+           [](vision::faceid::VPL& self, pybind11::array& data) {
+             auto mat = PyArrayToCvMat(data);
+             vision::FaceRecognitionResult res;
+             self.Predict(&mat, &res);
+             return res;
+           })
+      .def_readwrite("size", &vision::faceid::VPL::size)
+      .def_readwrite("alpha", &vision::faceid::VPL::alpha)
+      .def_readwrite("beta", &vision::faceid::VPL::beta)
+      .def_readwrite("swap_rb", &vision::faceid::VPL::swap_rb)
+      .def_readwrite("l2_normalize", &vision::faceid::VPL::l2_normalize);
+}
+
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/faceid/faceid_pybind.cc

@@ -0,0 +1,33 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+
+void BindArcFace(pybind11::module& m);
+void BindInsightFaceRecognitionModel(pybind11::module& m);
+void BindCosFace(pybind11::module& m);
+void BindPartialFC(pybind11::module& m);
+void BindVPL(pybind11::module& m);
+
+void BindFaceId(pybind11::module& m) {
+  auto faceid_module = m.def_submodule("faceid", "Face recognition models.");
+  BindInsightFaceRecognitionModel(faceid_module);
+  BindArcFace(faceid_module);
+  BindCosFace(faceid_module);
+  BindPartialFC(faceid_module);
+  BindVPL(faceid_module);
+}
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/matting/contrib/modnet.cc

@@ -12,7 +12,7 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-#include "fastdeploy/vision/zhkkke/modnet.h"
+#include "fastdeploy/vision/matting/contrib/modnet.h"
 #include "fastdeploy/utils/perf.h"
 #include "fastdeploy/vision/utils/utils.h"
 
@@ -20,7 +20,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace zhkkke {
+namespace matting {
 
 MODNet::MODNet(const std::string& model_file, const std::string& params_file,
                const RuntimeOption& custom_option,
@@ -170,6 +170,6 @@ bool MODNet::Predict(cv::Mat* im, MattingResult* result) {
   return true;
 }
 
-}  // namespace zhkkke
+}  // namespace matting
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/matting/contrib/modnet.h

@@ -21,7 +21,7 @@ namespace fastdeploy {
 
 namespace vision {
 
-namespace zhkkke {
+namespace matting {
 
 class FASTDEPLOY_DECL MODNet : public FastDeployModel {
  public:
@@ -32,7 +32,7 @@ class FASTDEPLOY_DECL MODNet : public FastDeployModel {
          const Frontend& model_format = Frontend::ONNX);
 
   // 定义模型的名称
-  std::string ModelName() const { return "ZHKKKe/MODNet"; }
+  std::string ModelName() const { return "matting/MODNet"; }
 
   // 以下为一些可供用户修改的属性
   // tuple of (width, height), default (256, 256)
@@ -65,6 +65,6 @@ class FASTDEPLOY_DECL MODNet : public FastDeployModel {
                    const std::map<std::string, std::array<int, 2>>& im_info);
 };
 
-}  // namespace zhkkke
+}  // namespace matting
 }  // namespace vision
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/matting/contrib/modnet_pybind.cc

@@ -15,23 +15,21 @@
 #include "fastdeploy/pybind/main.h"
 
 namespace fastdeploy {
-void BindZHKKKe(pybind11::module& m) {
+void BindMODNet(pybind11::module& m) {
-  auto zhkkke_module = m.def_submodule("zhkkke", "https://github.com/ZHKKKe");
   // Bind MODNet
-  pybind11::class_<vision::zhkkke::MODNet, FastDeployModel>(zhkkke_module,
+  pybind11::class_<vision::matting::MODNet, FastDeployModel>(m, "MODNet")
-                                                            "MODNet")
       .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
       .def("predict",
-           [](vision::zhkkke::MODNet& self, pybind11::array& data) {
+           [](vision::matting::MODNet& self, pybind11::array& data) {
              auto mat = PyArrayToCvMat(data);
              vision::MattingResult res;
              self.Predict(&mat, &res);
              return res;
            })
-      .def_readwrite("size", &vision::zhkkke::MODNet::size)
+      .def_readwrite("size", &vision::matting::MODNet::size)
-      .def_readwrite("alpha", &vision::zhkkke::MODNet::alpha)
+      .def_readwrite("alpha", &vision::matting::MODNet::alpha)
-      .def_readwrite("beta", &vision::zhkkke::MODNet::beta)
+      .def_readwrite("beta", &vision::matting::MODNet::beta)
-      .def_readwrite("swap_rb", &vision::zhkkke::MODNet::swap_rb);
+      .def_readwrite("swap_rb", &vision::matting::MODNet::swap_rb);
 }
 
 }  // namespace fastdeploy

csrcs/fastdeploy/vision/matting/matting_pybind.cc

@@ -0,0 +1,26 @@
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+// you may not use this file except in compliance with the License.
+// You may obtain a copy of the License at
+//
+//     http://www.apache.org/licenses/LICENSE-2.0
+//
+// Unless required by applicable law or agreed to in writing, software
+// distributed under the License is distributed on an "AS IS" BASIS,
+// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+// See the License for the specific language governing permissions and
+// limitations under the License.
+
+#include "fastdeploy/pybind/main.h"
+
+namespace fastdeploy {
+
+void BindMODNet(pybind11::module& m);
+
+void BindMatting(pybind11::module& m) {
+  auto matting_module =
+      m.def_submodule("matting", "Image object matting models.");
+  BindMODNet(matting_module);
+}
+}  // namespace fastdeploy

csrcs/fastdeploy/vision/ultralytics/ultralytics_pybind.cc

@@ -1,42 +0,0 @@
-// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
-//
-// Licensed under the Apache License, Version 2.0 (the "License");
-// you may not use this file except in compliance with the License.
-// You may obtain a copy of the License at
-//
-//     http://www.apache.org/licenses/LICENSE-2.0
-//
-// Unless required by applicable law or agreed to in writing, software
-// distributed under the License is distributed on an "AS IS" BASIS,
-// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-// See the License for the specific language governing permissions and
-// limitations under the License.
-
-#include "fastdeploy/pybind/main.h"
-
-namespace fastdeploy {
-void BindUltralytics(pybind11::module& m) {
-  auto ultralytics_module =
-      m.def_submodule("ultralytics", "https://github.com/ultralytics/yolov5");
-  pybind11::class_<vision::ultralytics::YOLOv5, FastDeployModel>(
-      ultralytics_module, "YOLOv5")
-      .def(pybind11::init<std::string, std::string, RuntimeOption, Frontend>())
-      .def("predict",
-           [](vision::ultralytics::YOLOv5& self, pybind11::array& data,
-              float conf_threshold, float nms_iou_threshold) {
-             auto mat = PyArrayToCvMat(data);
-             vision::DetectionResult res;
-             self.Predict(&mat, &res, conf_threshold, nms_iou_threshold);
-             return res;
-           })
-      .def_readwrite("size", &vision::ultralytics::YOLOv5::size)
-      .def_readwrite("padding_value",
-                     &vision::ultralytics::YOLOv5::padding_value)
-      .def_readwrite("is_mini_pad", &vision::ultralytics::YOLOv5::is_mini_pad)
-      .def_readwrite("is_no_pad", &vision::ultralytics::YOLOv5::is_no_pad)
-      .def_readwrite("is_scale_up", &vision::ultralytics::YOLOv5::is_scale_up)
-      .def_readwrite("stride", &vision::ultralytics::YOLOv5::stride)
-      .def_readwrite("max_wh", &vision::ultralytics::YOLOv5::max_wh)
-      .def_readwrite("multi_label", &vision::ultralytics::YOLOv5::multi_label);
-}
-}  // namespace fastdeploy

csrcs/fastdeploy/vision/vision_pybind.cc

@@ -19,18 +19,11 @@ namespace fastdeploy {
 void BindPPCls(pybind11::module& m);
 void BindPPDet(pybind11::module& m);
 void BindPPSeg(pybind11::module& m);
-void BindUltralytics(pybind11::module& m);
-void BindMeituan(pybind11::module& m);
-void BindMegvii(pybind11::module& m);
-void BindDeepCam(pybind11::module& m);
-void BindRangiLyu(pybind11::module& m);
-void BindLinzaer(pybind11::module& m);
-void BindBiubug6(pybind11::module& m);
-void BindPpogg(pybind11::module& m);
-void BindDeepInsight(pybind11::module& m);
-void BindZHKKKe(pybind11::module& m);
 
 void BindDetection(pybind11::module& m);
+void BindMatting(pybind11::module& m);
+void BindFaceDet(pybind11::module& m);
+void BindFaceId(pybind11::module& m);
 #ifdef ENABLE_VISION_VISUALIZE
 void BindVisualize(pybind11::module& m);
 #endif
@@ -88,18 +81,11 @@ void BindVision(pybind11::module& m) {
   BindPPCls(m);
   BindPPDet(m);
   BindPPSeg(m);
-  BindUltralytics(m);
-  BindMeituan(m);
-  BindMegvii(m);
-  BindDeepCam(m);
-  BindRangiLyu(m);
-  BindLinzaer(m);
-  BindBiubug6(m);
-  BindPpogg(m);
-  BindDeepInsight(m);
-  BindZHKKKe(m);
 
   BindDetection(m);
+  BindFaceDet(m);
+  BindFaceId(m);
+  BindMatting(m);
 #ifdef ENABLE_VISION_VISUALIZE
   BindVisualize(m);
 #endif

csrcs/fastdeploy/vision/visualize/visualize.h

@@ -26,10 +26,10 @@ class FASTDEPLOY_DECL Visualize {
   static std::vector<int> color_map_;
   static const std::vector<int>& GetColorMap(int num_classes = 1000);
   static cv::Mat VisDetection(const cv::Mat& im, const DetectionResult& result,
-                              int line_size = 2, float font_size = 0.5f);
+                              int line_size = 1, float font_size = 0.5f);
   static cv::Mat VisFaceDetection(const cv::Mat& im,
                                   const FaceDetectionResult& result,
-                                  int line_size = 2, float font_size = 0.5f);
+                                  int line_size = 1, float font_size = 0.5f);
   static cv::Mat VisSegmentation(const cv::Mat& im,
                                  const SegmentationResult& result);
   static cv::Mat VisMattingAlpha(const cv::Mat& im, const MattingResult& result,

fastdeploy/vision/__init__.py

@@ -13,21 +13,9 @@
 # limitations under the License.
 from __future__ import absolute_import
 
-from . import evaluation
-from . import ppcls
-from . import ppdet
-from . import ppseg
-from . import ultralytics
-from . import meituan
-from . import megvii
-from . import deepcam
-from . import rangilyu
-from . import linzaer
-from . import biubug6
-from . import ppogg
-from . import deepinsight
-from . import zhkkke
-
 from . import detection
+from . import matting
+from . import facedet
+from . import faceid
 
 from .visualize import *

fastdeploy/vision/deepinsight/__init__.py

@@ -1,563 +0,0 @@
-# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-from __future__ import absolute_import
-import logging
-from ... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
-
-
-class SCRFD(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(SCRFD, self).__init__(runtime_option)
-
-        self._model = C.vision.deepinsight.SCRFD(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "SCRFD initialize failed."
-
-    def predict(self, input_image, conf_threshold=0.7, nms_iou_threshold=0.3):
-        return self._model.predict(input_image, conf_threshold,
-                                   nms_iou_threshold)
-
-    # 一些跟SCRFD模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [640, 640]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def padding_value(self):
-        return self._model.padding_value
-
-    @property
-    def is_no_pad(self):
-        return self._model.is_no_pad
-
-    @property
-    def is_mini_pad(self):
-        return self._model.is_mini_pad
-
-    @property
-    def is_scale_up(self):
-        return self._model.is_scale_up
-
-    @property
-    def stride(self):
-        return self._model.stride
-
-    @property
-    def downsample_strides(self):
-        return self._model.downsample_strides
-
-    @property
-    def landmarks_per_face(self):
-        return self._model.landmarks_per_face
-
-    @property
-    def use_kps(self):
-        return self._model.use_kps
-
-    @property
-    def max_nms(self):
-        return self._model.max_nms
-
-    @property
-    def num_anchors(self):
-        return self._model.num_anchors
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @padding_value.setter
-    def padding_value(self, value):
-        assert isinstance(
-            value,
-            list), "The value to set `padding_value` must be type of list."
-        self._model.padding_value = value
-
-    @is_no_pad.setter
-    def is_no_pad(self, value):
-        assert isinstance(
-            value, bool), "The value to set `is_no_pad` must be type of bool."
-        self._model.is_no_pad = value
-
-    @is_mini_pad.setter
-    def is_mini_pad(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_mini_pad` must be type of bool."
-        self._model.is_mini_pad = value
-
-    @is_scale_up.setter
-    def is_scale_up(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_scale_up` must be type of bool."
-        self._model.is_scale_up = value
-
-    @stride.setter
-    def stride(self, value):
-        assert isinstance(
-            value, int), "The value to set `stride` must be type of int."
-        self._model.stride = value
-
-    @downsample_strides.setter
-    def downsample_strides(self, value):
-        assert isinstance(
-            value,
-            list), "The value to set `downsample_strides` must be type of list."
-        self._model.downsample_strides = value
-
-    @landmarks_per_face.setter
-    def landmarks_per_face(self, value):
-        assert isinstance(
-            value,
-            int), "The value to set `landmarks_per_face` must be type of int."
-        self._model.landmarks_per_face = value
-
-    @use_kps.setter
-    def use_kps(self, value):
-        assert isinstance(
-            value, bool), "The value to set `use_kps` must be type of bool."
-        self._model.use_kps = value
-
-    @max_nms.setter
-    def max_nms(self, value):
-        assert isinstance(
-            value, int), "The value to set `max_nms` must be type of int."
-        self._model.max_nms = value
-
-    @num_anchors.setter
-    def num_anchors(self, value):
-        assert isinstance(
-            value, int), "The value to set `num_anchors` must be type of int."
-        self._model.num_anchors = value
-
-
-class InsightFaceRecognitionModel(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(InsightFaceRecognitionModel, self).__init__(runtime_option)
-
-        self._model = C.vision.deepinsight.InsightFaceRecognitionModel(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "InsightFaceRecognitionModel initialize failed."
-
-    def predict(self, input_image):
-        return self._model.predict(input_image)
-
-    # 一些跟InsightFaceRecognitionModel模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def alpha(self):
-        return self._model.alpha
-
-    @property
-    def beta(self):
-        return self._model.beta
-
-    @property
-    def swap_rb(self):
-        return self._model.swap_rb
-
-    @property
-    def l2_normalize(self):
-        return self._model.l2_normalize
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @alpha.setter
-    def alpha(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `alpha` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.alpha = value
-
-    @beta.setter
-    def beta(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `beta` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.beta = value
-
-    @swap_rb.setter
-    def swap_rb(self, value):
-        assert isinstance(
-            value, bool), "The value to set `swap_rb` must be type of bool."
-        self._model.swap_rb = value
-
-    @l2_normalize.setter
-    def l2_normalize(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `l2_normalize` must be type of bool."
-        self._model.l2_normalize = value
-
-
-class ArcFace(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(ArcFace, self).__init__(runtime_option)
-
-        self._model = C.vision.deepinsight.ArcFace(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "ArcFace initialize failed."
-
-    def predict(self, input_image):
-        return self._model.predict(input_image)
-
-    # 一些跟模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def alpha(self):
-        return self._model.alpha
-
-    @property
-    def beta(self):
-        return self._model.beta
-
-    @property
-    def swap_rb(self):
-        return self._model.swap_rb
-
-    @property
-    def l2_normalize(self):
-        return self._model.l2_normalize
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @alpha.setter
-    def alpha(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `alpha` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.alpha = value
-
-    @beta.setter
-    def beta(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `beta` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.beta = value
-
-    @swap_rb.setter
-    def swap_rb(self, value):
-        assert isinstance(
-            value, bool), "The value to set `swap_rb` must be type of bool."
-        self._model.swap_rb = value
-
-    @l2_normalize.setter
-    def l2_normalize(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `l2_normalize` must be type of bool."
-        self._model.l2_normalize = value
-
-
-class CosFace(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(CosFace, self).__init__(runtime_option)
-
-        self._model = C.vision.deepinsight.CosFace(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "CosFace initialize failed."
-
-    def predict(self, input_image):
-        return self._model.predict(input_image)
-
-    # 一些跟模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def alpha(self):
-        return self._model.alpha
-
-    @property
-    def beta(self):
-        return self._model.beta
-
-    @property
-    def swap_rb(self):
-        return self._model.swap_rb
-
-    @property
-    def l2_normalize(self):
-        return self._model.l2_normalize
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @alpha.setter
-    def alpha(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `alpha` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.alpha = value
-
-    @beta.setter
-    def beta(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `beta` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.beta = value
-
-    @swap_rb.setter
-    def swap_rb(self, value):
-        assert isinstance(
-            value, bool), "The value to set `swap_rb` must be type of bool."
-        self._model.swap_rb = value
-
-    @l2_normalize.setter
-    def l2_normalize(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `l2_normalize` must be type of bool."
-        self._model.l2_normalize = value
-
-
-class PartialFC(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(PartialFC, self).__init__(runtime_option)
-
-        self._model = C.vision.deepinsight.PartialFC(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "PartialFC initialize failed."
-
-    def predict(self, input_image):
-        return self._model.predict(input_image)
-
-    # 一些跟模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def alpha(self):
-        return self._model.alpha
-
-    @property
-    def beta(self):
-        return self._model.beta
-
-    @property
-    def swap_rb(self):
-        return self._model.swap_rb
-
-    @property
-    def l2_normalize(self):
-        return self._model.l2_normalize
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @alpha.setter
-    def alpha(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `alpha` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.alpha = value
-
-    @beta.setter
-    def beta(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `beta` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.beta = value
-
-    @swap_rb.setter
-    def swap_rb(self, value):
-        assert isinstance(
-            value, bool), "The value to set `swap_rb` must be type of bool."
-        self._model.swap_rb = value
-
-    @l2_normalize.setter
-    def l2_normalize(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `l2_normalize` must be type of bool."
-        self._model.l2_normalize = value
-
-
-class VPL(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(VPL, self).__init__(runtime_option)
-
-        self._model = C.vision.deepinsight.VPL(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "VPL initialize failed."
-
-    def predict(self, input_image):
-        return self._model.predict(input_image)
-
-    # 一些跟模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def alpha(self):
-        return self._model.alpha
-
-    @property
-    def beta(self):
-        return self._model.beta
-
-    @property
-    def swap_rb(self):
-        return self._model.swap_rb
-
-    @property
-    def l2_normalize(self):
-        return self._model.l2_normalize
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @alpha.setter
-    def alpha(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `alpha` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.alpha = value
-
-    @beta.setter
-    def beta(self, value):
-        assert isinstance(value, (list, tuple)),\
-            "The value to set `beta` must be type of tuple or list."
-        assert len(value) == 3,\
-            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
-            len(value))
-        self._model.beta = value
-
-    @swap_rb.setter
-    def swap_rb(self, value):
-        assert isinstance(
-            value, bool), "The value to set `swap_rb` must be type of bool."
-        self._model.swap_rb = value
-
-    @l2_normalize.setter
-    def l2_normalize(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `l2_normalize` must be type of bool."
-        self._model.l2_normalize = value

fastdeploy/vision/detection/__init__.py

@@ -13,8 +13,12 @@
 # limitations under the License.
 
 from __future__ import absolute_import
-from .yolov7 import YOLOv7
+from .contrib.yolov7 import YOLOv7
-from .yolor import YOLOR
+from .contrib.yolor import YOLOR
-from .scaled_yolov4 import ScaledYOLOv4
+from .contrib.scaled_yolov4 import ScaledYOLOv4
-
+from .contrib.nanodet_plus import NanoDetPlus
+from .contrib.yolox import YOLOX
+from .contrib.yolov5 import YOLOv5
+from .contrib.yolov5lite import YOLOv5Lite
+from .contrib.yolov6 import YOLOv6
 from .ppdet import PPYOLOE, PPYOLO, PPYOLOv2, PaddleYOLOX, PicoDet, FasterRCNN, YOLOv3

fastdeploy/vision/detection/contrib/__init__.py

@@ -0,0 +1,15 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import

fastdeploy/vision/detection/contrib/nanodet_plus.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class NanoDetPlus(FastDeployModel):
@@ -28,7 +28,7 @@ class NanoDetPlus(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(NanoDetPlus, self).__init__(runtime_option)
 
-        self._model = C.vision.rangilyu.NanoDetPlus(
+        self._model = C.vision.detection.NanoDetPlus(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "NanoDetPlus initialize failed."

fastdeploy/vision/detection/contrib/scaled_yolov4.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class ScaledYOLOv4(FastDeployModel):

fastdeploy/vision/detection/contrib/yolor.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOR(FastDeployModel):

fastdeploy/vision/detection/contrib/yolov5.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOv5(FastDeployModel):
@@ -28,7 +28,7 @@ class YOLOv5(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(YOLOv5, self).__init__(runtime_option)
 
-        self._model = C.vision.ultralytics.YOLOv5(
+        self._model = C.vision.detection.YOLOv5(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "YOLOv5 initialize failed."

fastdeploy/vision/detection/contrib/yolov5lite.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOv5Lite(FastDeployModel):
@@ -28,7 +28,7 @@ class YOLOv5Lite(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(YOLOv5Lite, self).__init__(runtime_option)
 
-        self._model = C.vision.ppogg.YOLOv5Lite(
+        self._model = C.vision.detection.YOLOv5Lite(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "YOLOv5Lite initialize failed."

fastdeploy/vision/detection/contrib/yolov6.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOv6(FastDeployModel):
@@ -28,7 +28,7 @@ class YOLOv6(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(YOLOv6, self).__init__(runtime_option)
 
-        self._model = C.vision.meituan.YOLOv6(
+        self._model = C.vision.detection.YOLOv6(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "YOLOv6 initialize failed."

fastdeploy/vision/detection/contrib/yolov7.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOv7(FastDeployModel):

fastdeploy/vision/detection/contrib/yolox.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOX(FastDeployModel):
@@ -28,8 +28,8 @@ class YOLOX(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(YOLOX, self).__init__(runtime_option)
 
-        self._model = C.vision.megvii.YOLOX(model_file, params_file,
+        self._model = C.vision.detection.YOLOX(
-                                            self._runtime_option, model_format)
+            model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "YOLOX initialize failed."
 

fastdeploy/vision/facedet/__init__.py

@@ -0,0 +1,19 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+from .contrib.yolov5face import YOLOv5Face
+from .contrib.retinaface import RetinaFace
+from .contrib.scrfd import SCRFD
+from .contrib.ultraface import UltraFace

fastdeploy/vision/facedet/contrib/__init__.py

@@ -0,0 +1,15 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import

fastdeploy/vision/facedet/contrib/retinaface.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class RetinaFace(FastDeployModel):
@@ -28,7 +28,7 @@ class RetinaFace(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(RetinaFace, self).__init__(runtime_option)
 
-        self._model = C.vision.biubug6.RetinaFace(
+        self._model = C.vision.facedet.RetinaFace(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "RetinaFace initialize failed."

fastdeploy/vision/facedet/contrib/scrfd.py

@@ -0,0 +1,158 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+import logging
+from .... import FastDeployModel, Frontend
+from .... import c_lib_wrap as C
+
+
+class SCRFD(FastDeployModel):
+    def __init__(self,
+                 model_file,
+                 params_file="",
+                 runtime_option=None,
+                 model_format=Frontend.ONNX):
+        # 调用基函数进行backend_option的初始化
+        # 初始化后的option保存在self._runtime_option
+        super(SCRFD, self).__init__(runtime_option)
+
+        self._model = C.vision.facedet.SCRFD(
+            model_file, params_file, self._runtime_option, model_format)
+        # 通过self.initialized判断整个模型的初始化是否成功
+        assert self.initialized, "SCRFD initialize failed."
+
+    def predict(self, input_image, conf_threshold=0.7, nms_iou_threshold=0.3):
+        return self._model.predict(input_image, conf_threshold,
+                                   nms_iou_threshold)
+
+    # 一些跟SCRFD模型有关的属性封装
+    # 多数是预处理相关，可通过修改如model.size = [640, 640]改变预处理时resize的大小（前提是模型支持）
+    @property
+    def size(self):
+        return self._model.size
+
+    @property
+    def padding_value(self):
+        return self._model.padding_value
+
+    @property
+    def is_no_pad(self):
+        return self._model.is_no_pad
+
+    @property
+    def is_mini_pad(self):
+        return self._model.is_mini_pad
+
+    @property
+    def is_scale_up(self):
+        return self._model.is_scale_up
+
+    @property
+    def stride(self):
+        return self._model.stride
+
+    @property
+    def downsample_strides(self):
+        return self._model.downsample_strides
+
+    @property
+    def landmarks_per_face(self):
+        return self._model.landmarks_per_face
+
+    @property
+    def use_kps(self):
+        return self._model.use_kps
+
+    @property
+    def max_nms(self):
+        return self._model.max_nms
+
+    @property
+    def num_anchors(self):
+        return self._model.num_anchors
+
+    @size.setter
+    def size(self, wh):
+        assert isinstance(wh, (list, tuple)),\
+            "The value to set `size` must be type of tuple or list."
+        assert len(wh) == 2,\
+            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
+            len(wh))
+        self._model.size = wh
+
+    @padding_value.setter
+    def padding_value(self, value):
+        assert isinstance(
+            value,
+            list), "The value to set `padding_value` must be type of list."
+        self._model.padding_value = value
+
+    @is_no_pad.setter
+    def is_no_pad(self, value):
+        assert isinstance(
+            value, bool), "The value to set `is_no_pad` must be type of bool."
+        self._model.is_no_pad = value
+
+    @is_mini_pad.setter
+    def is_mini_pad(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `is_mini_pad` must be type of bool."
+        self._model.is_mini_pad = value
+
+    @is_scale_up.setter
+    def is_scale_up(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `is_scale_up` must be type of bool."
+        self._model.is_scale_up = value
+
+    @stride.setter
+    def stride(self, value):
+        assert isinstance(
+            value, int), "The value to set `stride` must be type of int."
+        self._model.stride = value
+
+    @downsample_strides.setter
+    def downsample_strides(self, value):
+        assert isinstance(
+            value,
+            list), "The value to set `downsample_strides` must be type of list."
+        self._model.downsample_strides = value
+
+    @landmarks_per_face.setter
+    def landmarks_per_face(self, value):
+        assert isinstance(
+            value,
+            int), "The value to set `landmarks_per_face` must be type of int."
+        self._model.landmarks_per_face = value
+
+    @use_kps.setter
+    def use_kps(self, value):
+        assert isinstance(
+            value, bool), "The value to set `use_kps` must be type of bool."
+        self._model.use_kps = value
+
+    @max_nms.setter
+    def max_nms(self, value):
+        assert isinstance(
+            value, int), "The value to set `max_nms` must be type of int."
+        self._model.max_nms = value
+
+    @num_anchors.setter
+    def num_anchors(self, value):
+        assert isinstance(
+            value, int), "The value to set `num_anchors` must be type of int."
+        self._model.num_anchors = value

fastdeploy/vision/facedet/contrib/ultraface.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class UltraFace(FastDeployModel):
@@ -28,7 +28,7 @@ class UltraFace(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(UltraFace, self).__init__(runtime_option)
 
-        self._model = C.vision.linzaer.UltraFace(
+        self._model = C.vision.facedet.UltraFace(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "UltraFace initialize failed."

fastdeploy/vision/facedet/contrib/yolov5face.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class YOLOv5Face(FastDeployModel):
@@ -28,7 +28,7 @@ class YOLOv5Face(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(YOLOv5Face, self).__init__(runtime_option)
 
-        self._model = C.vision.deepcam.YOLOv5Face(
+        self._model = C.vision.facedet.YOLOv5Face(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "YOLOv5Face initialize failed."

fastdeploy/vision/faceid/__init__.py

@@ -0,0 +1,20 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+from .contrib.arcface import ArcFace
+from .contrib.cosface import CosFace
+from .contrib.insightface_rec import InsightFaceRecognitionModel
+from .contrib.partial_fc import PartialFC
+from .contrib.vpl import VPL

fastdeploy/vision/faceid/contrib/__init__.py

@@ -0,0 +1,15 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import

fastdeploy/vision/faceid/contrib/arcface.py

@@ -0,0 +1,100 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+import logging
+from .... import FastDeployModel, Frontend
+from .... import c_lib_wrap as C
+from ..contrib.insightface_rec import InsightFaceRecognitionModel
+
+
+class ArcFace(FastDeployModel):
+    def __init__(self,
+                 model_file,
+                 params_file="",
+                 runtime_option=None,
+                 model_format=Frontend.ONNX):
+        # 调用基函数进行backend_option的初始化
+        # 初始化后的option保存在self._runtime_option
+        super(ArcFace, self).__init__(runtime_option)
+
+        self._model = C.vision.faceid.ArcFace(
+            model_file, params_file, self._runtime_option, model_format)
+        # 通过self.initialized判断整个模型的初始化是否成功
+        assert self.initialized, "ArcFace initialize failed."
+
+    def predict(self, input_image):
+        return self._model.predict(input_image)
+
+    # 一些跟模型有关的属性封装
+    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
+    @property
+    def size(self):
+        return self._model.size
+
+    @property
+    def alpha(self):
+        return self._model.alpha
+
+    @property
+    def beta(self):
+        return self._model.beta
+
+    @property
+    def swap_rb(self):
+        return self._model.swap_rb
+
+    @property
+    def l2_normalize(self):
+        return self._model.l2_normalize
+
+    @size.setter
+    def size(self, wh):
+        assert isinstance(wh, (list, tuple)),\
+            "The value to set `size` must be type of tuple or list."
+        assert len(wh) == 2,\
+            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
+            len(wh))
+        self._model.size = wh
+
+    @alpha.setter
+    def alpha(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `alpha` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.alpha = value
+
+    @beta.setter
+    def beta(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `beta` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.beta = value
+
+    @swap_rb.setter
+    def swap_rb(self, value):
+        assert isinstance(
+            value, bool), "The value to set `swap_rb` must be type of bool."
+        self._model.swap_rb = value
+
+    @l2_normalize.setter
+    def l2_normalize(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `l2_normalize` must be type of bool."
+        self._model.l2_normalize = value

fastdeploy/vision/faceid/contrib/cosface.py

@@ -0,0 +1,99 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+import logging
+from .... import FastDeployModel, Frontend
+from .... import c_lib_wrap as C
+
+
+class CosFace(FastDeployModel):
+    def __init__(self,
+                 model_file,
+                 params_file="",
+                 runtime_option=None,
+                 model_format=Frontend.ONNX):
+        # 调用基函数进行backend_option的初始化
+        # 初始化后的option保存在self._runtime_option
+        super(CosFace, self).__init__(runtime_option)
+
+        self._model = C.vision.faceid.CosFace(
+            model_file, params_file, self._runtime_option, model_format)
+        # 通过self.initialized判断整个模型的初始化是否成功
+        assert self.initialized, "CosFace initialize failed."
+
+    def predict(self, input_image):
+        return self._model.predict(input_image)
+
+    # 一些跟模型有关的属性封装
+    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
+    @property
+    def size(self):
+        return self._model.size
+
+    @property
+    def alpha(self):
+        return self._model.alpha
+
+    @property
+    def beta(self):
+        return self._model.beta
+
+    @property
+    def swap_rb(self):
+        return self._model.swap_rb
+
+    @property
+    def l2_normalize(self):
+        return self._model.l2_normalize
+
+    @size.setter
+    def size(self, wh):
+        assert isinstance(wh, (list, tuple)),\
+            "The value to set `size` must be type of tuple or list."
+        assert len(wh) == 2,\
+            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
+            len(wh))
+        self._model.size = wh
+
+    @alpha.setter
+    def alpha(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `alpha` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.alpha = value
+
+    @beta.setter
+    def beta(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `beta` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.beta = value
+
+    @swap_rb.setter
+    def swap_rb(self, value):
+        assert isinstance(
+            value, bool), "The value to set `swap_rb` must be type of bool."
+        self._model.swap_rb = value
+
+    @l2_normalize.setter
+    def l2_normalize(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `l2_normalize` must be type of bool."
+        self._model.l2_normalize = value

fastdeploy/vision/faceid/contrib/insightface_rec.py

@@ -0,0 +1,99 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+import logging
+from .... import FastDeployModel, Frontend
+from .... import c_lib_wrap as C
+
+
+class InsightFaceRecognitionModel(FastDeployModel):
+    def __init__(self,
+                 model_file,
+                 params_file="",
+                 runtime_option=None,
+                 model_format=Frontend.ONNX):
+        # 调用基函数进行backend_option的初始化
+        # 初始化后的option保存在self._runtime_option
+        super(InsightFaceRecognitionModel, self).__init__(runtime_option)
+
+        self._model = C.vision.faceid.InsightFaceRecognitionModel(
+            model_file, params_file, self._runtime_option, model_format)
+        # 通过self.initialized判断整个模型的初始化是否成功
+        assert self.initialized, "InsightFaceRecognitionModel initialize failed."
+
+    def predict(self, input_image):
+        return self._model.predict(input_image)
+
+    # 一些跟InsightFaceRecognitionModel模型有关的属性封装
+    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
+    @property
+    def size(self):
+        return self._model.size
+
+    @property
+    def alpha(self):
+        return self._model.alpha
+
+    @property
+    def beta(self):
+        return self._model.beta
+
+    @property
+    def swap_rb(self):
+        return self._model.swap_rb
+
+    @property
+    def l2_normalize(self):
+        return self._model.l2_normalize
+
+    @size.setter
+    def size(self, wh):
+        assert isinstance(wh, (list, tuple)),\
+            "The value to set `size` must be type of tuple or list."
+        assert len(wh) == 2,\
+            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
+            len(wh))
+        self._model.size = wh
+
+    @alpha.setter
+    def alpha(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `alpha` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.alpha = value
+
+    @beta.setter
+    def beta(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `beta` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.beta = value
+
+    @swap_rb.setter
+    def swap_rb(self, value):
+        assert isinstance(
+            value, bool), "The value to set `swap_rb` must be type of bool."
+        self._model.swap_rb = value
+
+    @l2_normalize.setter
+    def l2_normalize(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `l2_normalize` must be type of bool."
+        self._model.l2_normalize = value

fastdeploy/vision/faceid/contrib/partial_fc.py

@@ -0,0 +1,99 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+import logging
+from .... import FastDeployModel, Frontend
+from .... import c_lib_wrap as C
+
+
+class PartialFC(FastDeployModel):
+    def __init__(self,
+                 model_file,
+                 params_file="",
+                 runtime_option=None,
+                 model_format=Frontend.ONNX):
+        # 调用基函数进行backend_option的初始化
+        # 初始化后的option保存在self._runtime_option
+        super(PartialFC, self).__init__(runtime_option)
+
+        self._model = C.vision.faceid.PartialFC(
+            model_file, params_file, self._runtime_option, model_format)
+        # 通过self.initialized判断整个模型的初始化是否成功
+        assert self.initialized, "PartialFC initialize failed."
+
+    def predict(self, input_image):
+        return self._model.predict(input_image)
+
+    # 一些跟模型有关的属性封装
+    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
+    @property
+    def size(self):
+        return self._model.size
+
+    @property
+    def alpha(self):
+        return self._model.alpha
+
+    @property
+    def beta(self):
+        return self._model.beta
+
+    @property
+    def swap_rb(self):
+        return self._model.swap_rb
+
+    @property
+    def l2_normalize(self):
+        return self._model.l2_normalize
+
+    @size.setter
+    def size(self, wh):
+        assert isinstance(wh, (list, tuple)),\
+            "The value to set `size` must be type of tuple or list."
+        assert len(wh) == 2,\
+            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
+            len(wh))
+        self._model.size = wh
+
+    @alpha.setter
+    def alpha(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `alpha` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.alpha = value
+
+    @beta.setter
+    def beta(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `beta` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.beta = value
+
+    @swap_rb.setter
+    def swap_rb(self, value):
+        assert isinstance(
+            value, bool), "The value to set `swap_rb` must be type of bool."
+        self._model.swap_rb = value
+
+    @l2_normalize.setter
+    def l2_normalize(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `l2_normalize` must be type of bool."
+        self._model.l2_normalize = value

fastdeploy/vision/faceid/contrib/vpl.py

@@ -0,0 +1,99 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+import logging
+from .... import FastDeployModel, Frontend
+from .... import c_lib_wrap as C
+
+
+class VPL(FastDeployModel):
+    def __init__(self,
+                 model_file,
+                 params_file="",
+                 runtime_option=None,
+                 model_format=Frontend.ONNX):
+        # 调用基函数进行backend_option的初始化
+        # 初始化后的option保存在self._runtime_option
+        super(VPL, self).__init__(runtime_option)
+
+        self._model = C.vision.faceid.VPL(model_file, params_file,
+                                          self._runtime_option, model_format)
+        # 通过self.initialized判断整个模型的初始化是否成功
+        assert self.initialized, "VPL initialize failed."
+
+    def predict(self, input_image):
+        return self._model.predict(input_image)
+
+    # 一些跟模型有关的属性封装
+    # 多数是预处理相关，可通过修改如model.size = [112, 112]改变预处理时resize的大小（前提是模型支持）
+    @property
+    def size(self):
+        return self._model.size
+
+    @property
+    def alpha(self):
+        return self._model.alpha
+
+    @property
+    def beta(self):
+        return self._model.beta
+
+    @property
+    def swap_rb(self):
+        return self._model.swap_rb
+
+    @property
+    def l2_normalize(self):
+        return self._model.l2_normalize
+
+    @size.setter
+    def size(self, wh):
+        assert isinstance(wh, (list, tuple)),\
+            "The value to set `size` must be type of tuple or list."
+        assert len(wh) == 2,\
+            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
+            len(wh))
+        self._model.size = wh
+
+    @alpha.setter
+    def alpha(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `alpha` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `alpha` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.alpha = value
+
+    @beta.setter
+    def beta(self, value):
+        assert isinstance(value, (list, tuple)),\
+            "The value to set `beta` must be type of tuple or list."
+        assert len(value) == 3,\
+            "The value to set `beta` must contatins 3 elements for each channels, but now it contains {} elements.".format(
+            len(value))
+        self._model.beta = value
+
+    @swap_rb.setter
+    def swap_rb(self, value):
+        assert isinstance(
+            value, bool), "The value to set `swap_rb` must be type of bool."
+        self._model.swap_rb = value
+
+    @l2_normalize.setter
+    def l2_normalize(self, value):
+        assert isinstance(
+            value,
+            bool), "The value to set `l2_normalize` must be type of bool."
+        self._model.l2_normalize = value

fastdeploy/vision/matting/__init__.py

@@ -0,0 +1,16 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import
+from .contrib.modnet import MODNet

fastdeploy/vision/matting/contrib/__init__.py

@@ -0,0 +1,15 @@
+# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+from __future__ import absolute_import

fastdeploy/vision/matting/contrib/modnet.py

@@ -14,8 +14,8 @@
 
 from __future__ import absolute_import
 import logging
-from ... import FastDeployModel, Frontend
+from .... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
+from .... import c_lib_wrap as C
 
 
 class MODNet(FastDeployModel):
@@ -28,7 +28,7 @@ class MODNet(FastDeployModel):
         # 初始化后的option保存在self._runtime_option
         super(MODNet, self).__init__(runtime_option)
 
-        self._model = C.vision.zhkkke.MODNet(
+        self._model = C.vision.matting.MODNet(
             model_file, params_file, self._runtime_option, model_format)
         # 通过self.initialized判断整个模型的初始化是否成功
         assert self.initialized, "MODNet initialize failed."

fastdeploy/vision/visualize/__init__.py

@@ -17,12 +17,12 @@ import logging
 from ... import c_lib_wrap as C
 
 
-def vis_detection(im_data, det_result, line_size=2, font_size=0.5):
+def vis_detection(im_data, det_result, line_size=1, font_size=0.5):
     return C.vision.Visualize.vis_detection(im_data, det_result, line_size,
                                             font_size)
 
 
-def vis_face_detection(im_data, face_det_result, line_size=2, font_size=0.5):
+def vis_face_detection(im_data, face_det_result, line_size=1, font_size=0.5):
     return C.vision.Visualize.vis_face_detection(im_data, face_det_result,
                                                  line_size, font_size)
 

fastdeploy/vision/wongkinyiu/__init__.py

@@ -1,312 +0,0 @@
-# Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.
-#
-# Licensed under the Apache License, Version 2.0 (the "License");
-# you may not use this file except in compliance with the License.
-# You may obtain a copy of the License at
-#
-#     http://www.apache.org/licenses/LICENSE-2.0
-#
-# Unless required by applicable law or agreed to in writing, software
-# distributed under the License is distributed on an "AS IS" BASIS,
-# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-# See the License for the specific language governing permissions and
-# limitations under the License.
-
-from __future__ import absolute_import
-import logging
-from ... import FastDeployModel, Frontend
-from ... import c_lib_wrap as C
-
-
-class YOLOv7(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(YOLOv7, self).__init__(runtime_option)
-
-        self._model = C.vision.wongkinyiu.YOLOv7(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "YOLOv7 initialize failed."
-
-    def predict(self, input_image, conf_threshold=0.25, nms_iou_threshold=0.5):
-        return self._model.predict(input_image, conf_threshold,
-                                   nms_iou_threshold)
-
-    # 一些跟YOLOv7模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [1280, 1280]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def padding_value(self):
-        return self._model.padding_value
-
-    @property
-    def is_no_pad(self):
-        return self._model.is_no_pad
-
-    @property
-    def is_mini_pad(self):
-        return self._model.is_mini_pad
-
-    @property
-    def is_scale_up(self):
-        return self._model.is_scale_up
-
-    @property
-    def stride(self):
-        return self._model.stride
-
-    @property
-    def max_wh(self):
-        return self._model.max_wh
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @padding_value.setter
-    def padding_value(self, value):
-        assert isinstance(
-            value,
-            list), "The value to set `padding_value` must be type of list."
-        self._model.padding_value = value
-
-    @is_no_pad.setter
-    def is_no_pad(self, value):
-        assert isinstance(
-            value, bool), "The value to set `is_no_pad` must be type of bool."
-        self._model.is_no_pad = value
-
-    @is_mini_pad.setter
-    def is_mini_pad(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_mini_pad` must be type of bool."
-        self._model.is_mini_pad = value
-
-    @is_scale_up.setter
-    def is_scale_up(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_scale_up` must be type of bool."
-        self._model.is_scale_up = value
-
-    @stride.setter
-    def stride(self, value):
-        assert isinstance(
-            value, int), "The value to set `stride` must be type of int."
-        self._model.stride = value
-
-    @max_wh.setter
-    def max_wh(self, value):
-        assert isinstance(
-            value, float), "The value to set `max_wh` must be type of float."
-        self._model.max_wh = value
-
-
-class YOLOR(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(YOLOR, self).__init__(runtime_option)
-
-        self._model = C.vision.wongkinyiu.YOLOR(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "YOLOR initialize failed."
-
-    def predict(self, input_image, conf_threshold=0.25, nms_iou_threshold=0.5):
-        return self._model.predict(input_image, conf_threshold,
-                                   nms_iou_threshold)
-
-    # 一些跟YOLOR模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [1280, 1280]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def padding_value(self):
-        return self._model.padding_value
-
-    @property
-    def is_no_pad(self):
-        return self._model.is_no_pad
-
-    @property
-    def is_mini_pad(self):
-        return self._model.is_mini_pad
-
-    @property
-    def is_scale_up(self):
-        return self._model.is_scale_up
-
-    @property
-    def stride(self):
-        return self._model.stride
-
-    @property
-    def max_wh(self):
-        return self._model.max_wh
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @padding_value.setter
-    def padding_value(self, value):
-        assert isinstance(
-            value,
-            list), "The value to set `padding_value` must be type of list."
-        self._model.padding_value = value
-
-    @is_no_pad.setter
-    def is_no_pad(self, value):
-        assert isinstance(
-            value, bool), "The value to set `is_no_pad` must be type of bool."
-        self._model.is_no_pad = value
-
-    @is_mini_pad.setter
-    def is_mini_pad(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_mini_pad` must be type of bool."
-        self._model.is_mini_pad = value
-
-    @is_scale_up.setter
-    def is_scale_up(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_scale_up` must be type of bool."
-        self._model.is_scale_up = value
-
-    @stride.setter
-    def stride(self, value):
-        assert isinstance(
-            value, int), "The value to set `stride` must be type of int."
-        self._model.stride = value
-
-    @max_wh.setter
-    def max_wh(self, value):
-        assert isinstance(
-            value, float), "The value to set `max_wh` must be type of float."
-        self._model.max_wh = value
-
-
-class ScaledYOLOv4(FastDeployModel):
-    def __init__(self,
-                 model_file,
-                 params_file="",
-                 runtime_option=None,
-                 model_format=Frontend.ONNX):
-        # 调用基函数进行backend_option的初始化
-        # 初始化后的option保存在self._runtime_option
-        super(ScaledYOLOv4, self).__init__(runtime_option)
-
-        self._model = C.vision.wongkinyiu.ScaledYOLOv4(
-            model_file, params_file, self._runtime_option, model_format)
-        # 通过self.initialized判断整个模型的初始化是否成功
-        assert self.initialized, "ScaledYOLOv4 initialize failed."
-
-    def predict(self, input_image, conf_threshold=0.25, nms_iou_threshold=0.5):
-        return self._model.predict(input_image, conf_threshold,
-                                   nms_iou_threshold)
-
-    # 一些跟ScaledYOLOv4模型有关的属性封装
-    # 多数是预处理相关，可通过修改如model.size = [1280, 1280]改变预处理时resize的大小（前提是模型支持）
-    @property
-    def size(self):
-        return self._model.size
-
-    @property
-    def padding_value(self):
-        return self._model.padding_value
-
-    @property
-    def is_no_pad(self):
-        return self._model.is_no_pad
-
-    @property
-    def is_mini_pad(self):
-        return self._model.is_mini_pad
-
-    @property
-    def is_scale_up(self):
-        return self._model.is_scale_up
-
-    @property
-    def stride(self):
-        return self._model.stride
-
-    @property
-    def max_wh(self):
-        return self._model.max_wh
-
-    @size.setter
-    def size(self, wh):
-        assert isinstance(wh, (list, tuple)),\
-            "The value to set `size` must be type of tuple or list."
-        assert len(wh) == 2,\
-            "The value to set `size` must contatins 2 elements means [width, height], but now it contains {} elements.".format(
-            len(wh))
-        self._model.size = wh
-
-    @padding_value.setter
-    def padding_value(self, value):
-        assert isinstance(
-            value,
-            list), "The value to set `padding_value` must be type of list."
-        self._model.padding_value = value
-
-    @is_no_pad.setter
-    def is_no_pad(self, value):
-        assert isinstance(
-            value, bool), "The value to set `is_no_pad` must be type of bool."
-        self._model.is_no_pad = value
-
-    @is_mini_pad.setter
-    def is_mini_pad(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_mini_pad` must be type of bool."
-        self._model.is_mini_pad = value
-
-    @is_scale_up.setter
-    def is_scale_up(self, value):
-        assert isinstance(
-            value,
-            bool), "The value to set `is_scale_up` must be type of bool."
-        self._model.is_scale_up = value
-
-    @stride.setter
-    def stride(self, value):
-        assert isinstance(
-            value, int), "The value to set `stride` must be type of int."
-        self._model.stride = value
-
-    @max_wh.setter
-    def max_wh(self, value):
-        assert isinstance(
-            value, float), "The value to set `max_wh` must be type of float."
-        self._model.max_wh = value

new_examples/vision/detection/yolov7/cpp/README.md

@@ -35,7 +35,7 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000
 ### YOLOv7类
 
 ```
-fastdeploy::vision::wongkinyiu::YOLOv7(
+fastdeploy::vision::detection::YOLOv7(
         const string& model_file,
         const string& params_file = "",
         const RuntimeOption& runtime_option = RuntimeOption(),

new_examples/vision/detection/yolov7/cpp/infer.cc

@@ -35,7 +35,7 @@ void CpuInfer(const std::string& model_file, const std::string& image_file) {
   std::cout << "Visualized result saved in ./vis_result.jpg" << std::endl;
 }
 
-void GpuInfer() {
+void GpuInfer(const std::string& model_file, const std::string& image_file) {
   auto option = fastdeploy::RuntimeOption();
   option.UseGpu();
   auto model = fastdeploy::vision::detection::YOLOv7(model_file, "", option);
@@ -58,12 +58,11 @@ void GpuInfer() {
   std::cout << "Visualized result saved in ./vis_result.jpg" << std::endl;
 }
 
-void TrtInfer() {
+void TrtInfer(const std::string& model_file, const std::string& image_file) {
   auto option = fastdeploy::RuntimeOption();
   option.UseGpu();
   option.UseTrtBackend();
-  option.SetTrtInputShape("image", {1, 3, 320, 320}, {1, 3, 640, 640},
+  option.SetTrtInputShape("images", {1, 3, 640, 640});
-                          {1, 3, 1280, 1280});
   auto model = fastdeploy::vision::detection::YOLOv7(model_file, "", option);
   if (!model.Initialized()) {
     std::cerr << "Failed to initialize." << std::endl;
@@ -96,11 +95,11 @@ int main(int argc, char* argv[]) {
   }
 
   if (std::atoi(argv[3]) == 0) {
-    CpuInfer();
+    CpuInfer(argv[1], argv[2]);
   } else if (std::atoi(argv[3]) == 1) {
-    GpuInfer();
+    GpuInfer(argv[1], argv[2]);
   } else if (std::atoi(argv[3]) == 2) {
-    TrtInfer();
+    TrtInfer(argv[1], argv[2]);
   }
   return 0;
 }

new_examples/vision/detection/yolov7/python/README.md

@@ -10,7 +10,8 @@
 ```
 #下载yolov7模型文件和测试图片
 wget https://bj.bcebos.com/paddlehub/fastdeploy/yolov7.onnx
-wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000087038.jpg
+wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg
+
 
 #下载部署示例代码
 git clone https://github.com/PaddlePaddle/FastDeploy.git

new_examples/vision/detection/yolov7/python/infer.py

@@ -31,8 +31,7 @@ def build_option(args):
 
     if args.use_trt:
         option.use_trt_backend()
-        option.set_trt_input_shape("image", [1, 3, 320, 320],
+        option.set_trt_input_shape("images", [1, 3, 640, 640])
-                                   [1, 3, 640, 640], [1, 3, 1280, 1280])
     return option