Add RetinaFace Model support (#48)

* 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 * Add RetinaFace Model support * fixed retinaface/api.md typos
2025-12-24 13:28:13 +08:00 · 2022-07-28 10:35:27 +08:00
parent 841302ceab
commit adddd3c452
18 changed files with 911 additions and 11 deletions
--- a/csrcs/fastdeploy/vision.h
+++ b/csrcs/fastdeploy/vision.h
@@ -15,6 +15,7 @@

 #include "fastdeploy/core/config.h"
 #ifdef ENABLE_VISION
+#include "fastdeploy/vision/biubug6/retinaface.h"
 #include "fastdeploy/vision/deepcam/yolov5face.h"
 #include "fastdeploy/vision/linzaer/ultraface.h"
 #include "fastdeploy/vision/megvii/yolox.h"
--- a/csrcs/fastdeploy/vision/biubug6/biubug6_pybind.cc
+++ b/csrcs/fastdeploy/vision/biubug6/biubug6_pybind.cc
@@ -0,0 +1,40 @@
+// 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 BindBiubug6(pybind11::module& m) {
+  auto biubug6_module = m.def_submodule(
+      "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,
+              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("variance", &vision::biubug6::RetinaFace::variance)
+      .def_readwrite("downsample_strides",
+                     &vision::biubug6::RetinaFace::downsample_strides)
+      .def_readwrite("min_sizes", &vision::biubug6::RetinaFace::min_sizes)
+      .def_readwrite("landmarks_per_face",
+                     &vision::biubug6::RetinaFace::landmarks_per_face);
+}
+}  // namespace fastdeploy
--- a/csrcs/fastdeploy/vision/biubug6/retinaface.cc
+++ b/csrcs/fastdeploy/vision/biubug6/retinaface.cc
@@ -0,0 +1,310 @@
+// 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/vision/biubug6/retinaface.h"
+#include "fastdeploy/utils/perf.h"
+#include "fastdeploy/vision/utils/utils.h"
+
+namespace fastdeploy {
+
+namespace vision {
+
+namespace biubug6 {
+
+struct RetinaAnchor {
+  float cx;
+  float cy;
+  float s_kx;
+  float s_ky;
+};
+
+void GenerateRetinaAnchors(const std::vector<int>& size,
+                           const std::vector<int>& downsample_strides,
+                           const std::vector<std::vector<int>>& min_sizes,
+                           std::vector<RetinaAnchor>* anchors) {
+  // size: tuple of input (width, height)
+  // downsample_strides: downsample strides (steps), e.g (8,16,32)
+  // min_sizes: width and height for each anchor,
+  // e.g {{16, 32}, {64, 128}, {256, 512}}
+  int h = size[1];
+  int w = size[0];
+  std::vector<std::vector<int>> feature_maps;
+  for (auto s : downsample_strides) {
+    feature_maps.push_back(
+        {static_cast<int>(
+             std::ceil(static_cast<float>(h) / static_cast<float>(s))),
+         static_cast<int>(
+             std::ceil(static_cast<float>(w) / static_cast<float>(s)))});
+  }
+
+  (*anchors).clear();
+  const size_t num_feature_map = feature_maps.size();
+  // reference: layers/functions/prior_box.py#L21
+  for (size_t k = 0; k < num_feature_map; ++k) {
+    auto f_map = feature_maps.at(k);       // e.g [640//8,640//8]
+    auto tmp_min_sizes = min_sizes.at(k);  // e.g [8,16]
+    int f_h = f_map.at(0);
+    int f_w = f_map.at(1);
+    for (size_t i = 0; i < f_h; ++i) {
+      for (size_t j = 0; j < f_w; ++j) {
+        for (auto min_size : tmp_min_sizes) {
+          float s_kx =
+              static_cast<float>(min_size) / static_cast<float>(w);  // e.g 16/w
+          float s_ky =
+              static_cast<float>(min_size) / static_cast<float>(h);  // e.g 16/h
+          // (x + 0.5) * step / w normalized loc mapping to input width
+          // (y + 0.5) * step / h normalized loc mapping to input height
+          float s = static_cast<float>(downsample_strides.at(k));
+          float cx = (static_cast<float>(j) + 0.5f) * s / static_cast<float>(w);
+          float cy = (static_cast<float>(i) + 0.5f) * s / static_cast<float>(h);
+          (*anchors).emplace_back(
+              RetinaAnchor{cx, cy, s_kx, s_ky});  // without clip
+        }
+      }
+    }
+  }
+}
+
+RetinaFace::RetinaFace(const std::string& model_file,
+                       const std::string& params_file,
+                       const RuntimeOption& custom_option,
+                       const Frontend& model_format) {
+  if (model_format == Frontend::ONNX) {
+    valid_cpu_backends = {Backend::ORT};  // 指定可用的CPU后端
+    valid_gpu_backends = {Backend::ORT, Backend::TRT};  // 指定可用的GPU后端
+  } else {
+    valid_cpu_backends = {Backend::PDINFER, Backend::ORT};
+    valid_gpu_backends = {Backend::PDINFER, Backend::ORT, Backend::TRT};
+  }
+  runtime_option = custom_option;
+  runtime_option.model_format = model_format;
+  runtime_option.model_file = model_file;
+  runtime_option.params_file = params_file;
+  initialized = Initialize();
+}
+
+bool RetinaFace::Initialize() {
+  // parameters for preprocess
+  size = {640, 640};
+  variance = {0.1f, 0.2f};
+  downsample_strides = {8, 16, 32};
+  min_sizes = {{16, 32}, {64, 128}, {256, 512}};
+  landmarks_per_face = 5;
+
+  if (!InitRuntime()) {
+    FDERROR << "Failed to initialize fastdeploy backend." << std::endl;
+    return false;
+  }
+  // Check if the input shape is dynamic after Runtime already initialized,
+  is_dynamic_input_ = false;
+  auto shape = InputInfoOfRuntime(0).shape;
+  for (int i = 0; i < shape.size(); ++i) {
+    // if height or width is dynamic
+    if (i >= 2 && shape[i] <= 0) {
+      is_dynamic_input_ = true;
+      break;
+    }
+  }
+  return true;
+}
+
+bool RetinaFace::Preprocess(
+    Mat* mat, FDTensor* output,
+    std::map<std::string, std::array<float, 2>>* im_info) {
+  // retinaface's preprocess steps
+  // 1. Resize
+  // 2. Convert(opencv style) or Normalize
+  // 3. HWC->CHW
+  int resize_w = size[0];
+  int resize_h = size[1];
+  if (resize_h != mat->Height() || resize_w != mat->Width()) {
+    Resize::Run(mat, resize_w, resize_h);
+  }
+
+  // Compute `result = mat * alpha + beta` directly by channel
+  // Reference: detect.py#L94
+  std::vector<float> alpha = {1.f, 1.f, 1.f};
+  std::vector<float> beta = {-104.f, -117.f, -123.f};  // BGR;
+  Convert::Run(mat, alpha, beta);
+
+  // Record output shape of preprocessed image
+  (*im_info)["output_shape"] = {static_cast<float>(mat->Height()),
+                                static_cast<float>(mat->Width())};
+
+  HWC2CHW::Run(mat);
+  Cast::Run(mat, "float");
+  mat->ShareWithTensor(output);
+  output->shape.insert(output->shape.begin(), 1);  // reshape to n, h, w, c
+  return true;
+}
+
+bool RetinaFace::Postprocess(
+    std::vector<FDTensor>& infer_result, FaceDetectionResult* result,
+    const std::map<std::string, std::array<float, 2>>& im_info,
+    float conf_threshold, float nms_iou_threshold) {
+  // retinaface has 3 output tensors, boxes & conf & landmarks
+  FDASSERT(
+      (infer_result.size() == 3),
+      "The default number of output tensor must be 3 according to retinaface.");
+  FDTensor& boxes_tensor = infer_result.at(0);      // (1,n,4)
+  FDTensor& conf_tensor = infer_result.at(1);       // (1,n,2)
+  FDTensor& landmarks_tensor = infer_result.at(2);  // (1,n,10)
+  FDASSERT((boxes_tensor.shape[0] == 1), "Only support batch =1 now.");
+  if (boxes_tensor.dtype != FDDataType::FP32) {
+    FDERROR << "Only support post process with float32 data." << std::endl;
+    return false;
+  }
+
+  result->Clear();
+  // must be setup landmarks_per_face before reserve
+  result->landmarks_per_face = landmarks_per_face;
+  result->Reserve(boxes_tensor.shape[1]);
+
+  float* boxes_ptr = static_cast<float*>(boxes_tensor.Data());
+  float* conf_ptr = static_cast<float*>(conf_tensor.Data());
+  float* landmarks_ptr = static_cast<float*>(landmarks_tensor.Data());
+  const size_t num_bboxes = boxes_tensor.shape[1];  // n
+  // fetch original image shape
+  auto iter_ipt = im_info.find("input_shape");
+  FDASSERT((iter_ipt != im_info.end()),
+           "Cannot find input_shape from im_info.");
+  float ipt_h = iter_ipt->second[0];
+  float ipt_w = iter_ipt->second[1];
+
+  // generate anchors with dowmsample strides
+  std::vector<RetinaAnchor> anchors;
+  GenerateRetinaAnchors(size, downsample_strides, min_sizes, &anchors);
+
+  // decode bounding boxes
+  for (size_t i = 0; i < num_bboxes; ++i) {
+    float confidence = conf_ptr[2 * i + 1];
+    // filter boxes by conf_threshold
+    if (confidence <= conf_threshold) {
+      continue;
+    }
+    float prior_cx = anchors.at(i).cx;
+    float prior_cy = anchors.at(i).cy;
+    float prior_s_kx = anchors.at(i).s_kx;
+    float prior_s_ky = anchors.at(i).s_ky;
+
+    // fetch offsets (dx,dy,dw,dh)
+    float dx = boxes_ptr[4 * i + 0];
+    float dy = boxes_ptr[4 * i + 1];
+    float dw = boxes_ptr[4 * i + 2];
+    float dh = boxes_ptr[4 * i + 3];
+    // reference: Pytorch_Retinaface/utils/box_utils.py
+    float x = prior_cx + dx * variance[0] * prior_s_kx;
+    float y = prior_cy + dy * variance[0] * prior_s_ky;
+    float w = prior_s_kx * std::exp(dw * variance[1]);
+    float h = prior_s_ky * std::exp(dh * variance[1]);  // (0.~1.)
+    // from (x,y,w,h) to (x1,y1,x2,y2)
+    float x1 = (x - w / 2.f) * ipt_w;
+    float y1 = (y - h / 2.f) * ipt_h;
+    float x2 = (x + w / 2.f) * ipt_w;
+    float y2 = (y + h / 2.f) * ipt_h;
+    result->boxes.emplace_back(std::array<float, 4>{x1, y1, x2, y2});
+    result->scores.push_back(confidence);
+    // decode landmarks (default 5 landmarks)
+    if (landmarks_per_face > 0) {
+      // reference: utils/box_utils.py#L241
+      for (size_t j = 0; j < landmarks_per_face * 2; j += 2) {
+        float ldx = landmarks_ptr[i * (landmarks_per_face * 2) + (j + 0)];
+        float ldy = landmarks_ptr[i * (landmarks_per_face * 2) + (j + 1)];
+        float lx = (prior_cx + ldx * variance[0] * prior_s_kx) * ipt_w;
+        float ly = (prior_cy + ldy * variance[0] * prior_s_ky) * ipt_h;
+        result->landmarks.emplace_back(std::array<float, 2>{lx, ly});
+      }
+    }
+  }
+
+  if (result->boxes.size() == 0) {
+    return true;
+  }
+
+  utils::NMS(result, nms_iou_threshold);
+
+  // scale and clip box
+  for (size_t i = 0; i < result->boxes.size(); ++i) {
+    result->boxes[i][0] = std::max(result->boxes[i][0], 0.0f);
+    result->boxes[i][1] = std::max(result->boxes[i][1], 0.0f);
+    result->boxes[i][2] = std::max(result->boxes[i][2], 0.0f);
+    result->boxes[i][3] = std::max(result->boxes[i][3], 0.0f);
+    result->boxes[i][0] = std::min(result->boxes[i][0], ipt_w - 1.0f);
+    result->boxes[i][1] = std::min(result->boxes[i][1], ipt_h - 1.0f);
+    result->boxes[i][2] = std::min(result->boxes[i][2], ipt_w - 1.0f);
+    result->boxes[i][3] = std::min(result->boxes[i][3], ipt_h - 1.0f);
+  }
+  // scale and clip landmarks
+  for (size_t i = 0; i < result->landmarks.size(); ++i) {
+    result->landmarks[i][0] = std::max(result->landmarks[i][0], 0.0f);
+    result->landmarks[i][1] = std::max(result->landmarks[i][1], 0.0f);
+    result->landmarks[i][0] = std::min(result->landmarks[i][0], ipt_w - 1.0f);
+    result->landmarks[i][1] = std::min(result->landmarks[i][1], ipt_h - 1.0f);
+  }
+  return true;
+}
+
+bool RetinaFace::Predict(cv::Mat* im, FaceDetectionResult* result,
+                         float conf_threshold, float nms_iou_threshold) {
+#ifdef FASTDEPLOY_DEBUG
+  TIMERECORD_START(0)
+#endif
+
+  Mat mat(*im);
+  std::vector<FDTensor> input_tensors(1);
+
+  std::map<std::string, std::array<float, 2>> im_info;
+
+  // Record the shape of image and the shape of preprocessed image
+  im_info["input_shape"] = {static_cast<float>(mat.Height()),
+                            static_cast<float>(mat.Width())};
+  im_info["output_shape"] = {static_cast<float>(mat.Height()),
+                             static_cast<float>(mat.Width())};
+
+  if (!Preprocess(&mat, &input_tensors[0], &im_info)) {
+    FDERROR << "Failed to preprocess input image." << std::endl;
+    return false;
+  }
+
+#ifdef FASTDEPLOY_DEBUG
+  TIMERECORD_END(0, "Preprocess")
+  TIMERECORD_START(1)
+#endif
+
+  input_tensors[0].name = InputInfoOfRuntime(0).name;
+  std::vector<FDTensor> output_tensors;
+  if (!Infer(input_tensors, &output_tensors)) {
+    FDERROR << "Failed to inference." << std::endl;
+    return false;
+  }
+#ifdef FASTDEPLOY_DEBUG
+  TIMERECORD_END(1, "Inference")
+  TIMERECORD_START(2)
+#endif
+
+  if (!Postprocess(output_tensors, result, im_info, conf_threshold,
+                   nms_iou_threshold)) {
+    FDERROR << "Failed to post process." << std::endl;
+    return false;
+  }
+
+#ifdef FASTDEPLOY_DEBUG
+  TIMERECORD_END(2, "Postprocess")
+#endif
+  return true;
+}
+
+}  // namespace biubug6
+}  // namespace vision
+}  // namespace fastdeploy
--- a/csrcs/fastdeploy/vision/biubug6/retinaface.h
+++ b/csrcs/fastdeploy/vision/biubug6/retinaface.h
@@ -0,0 +1,92 @@
+// 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.
+
+#pragma once
+#include "fastdeploy/fastdeploy_model.h"
+#include "fastdeploy/vision/common/processors/transform.h"
+#include "fastdeploy/vision/common/result.h"
+
+namespace fastdeploy {
+
+namespace vision {
+
+namespace biubug6 {
+
+class FASTDEPLOY_DECL RetinaFace : public FastDeployModel {
+ public:
+  // 当model_format为ONNX时，无需指定params_file
+  // 当model_format为Paddle时，则需同时指定model_file & params_file
+  RetinaFace(const std::string& model_file, const std::string& params_file = "",
+             const RuntimeOption& custom_option = RuntimeOption(),
+             const Frontend& model_format = Frontend::ONNX);
+
+  // 定义模型的名称
+  std::string ModelName() const { return "biubug6/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
+  // result 为模型预测的输出结构体
+  // conf_threshold 为后处理的参数
+  // nms_iou_threshold 为后处理的参数
+  virtual bool Predict(cv::Mat* im, FaceDetectionResult* result,
+                       float conf_threshold = 0.25f,
+                       float nms_iou_threshold = 0.4f);
+
+  // 以下为模型在预测时的一些参数，基本是前后处理所需
+  // 用户在创建模型后，可根据模型的要求，以及自己的需求
+  // 对参数进行修改
+  // tuple of (width, height), default (640, 640)
+  std::vector<int> size;
+  // variance in RetinaFace's prior-box(anchor) generate process,
+  // default (0.1, 0.2)
+  std::vector<float> variance;
+  // downsample strides (namely, steps) for RetinaFace to
+  // generate anchors, will take (8,16,32) as default values.
+  std::vector<int> downsample_strides;
+  // min sizes, width and height for each anchor.
+  std::vector<std::vector<int>> min_sizes;
+  // landmarks_per_face, default 5 in RetinaFace
+  int landmarks_per_face;
+
+ private:
+  // 初始化函数，包括初始化后端，以及其它模型推理需要涉及的操作
+  bool Initialize();
+
+  // 输入图像预处理操作
+  // Mat为FastDeploy定义的数据结构
+  // FDTensor为预处理后的Tensor数据，传给后端进行推理
+  // im_info为预处理过程保存的数据，在后处理中需要用到
+  bool Preprocess(Mat* mat, FDTensor* output,
+                  std::map<std::string, std::array<float, 2>>* im_info);
+
+  // 后端推理结果后处理，输出给用户
+  // infer_result 为后端推理后的输出Tensor
+  // result 为模型预测的结果
+  // 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(std::vector<FDTensor>& infer_result,
+                   FaceDetectionResult* 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_; }
+
+  bool is_dynamic_input_;
+};
+
+}  // namespace biubug6
+}  // namespace vision
+}  // namespace fastdeploy
--- a/csrcs/fastdeploy/vision/deepcam/yolov5face.cc
+++ b/csrcs/fastdeploy/vision/deepcam/yolov5face.cc
@@ -155,14 +155,16 @@ bool YOLOv5Face::Postprocess(
    float conf_threshold, float nms_iou_threshold) {
  // infer_result: (1,n,16) 16=4+1+10+1
  FDASSERT(infer_result.shape[0] == 1, "Only support batch =1 now.");
-  result->Clear();
-  // must be setup landmarks_per_face before reserve
-  result->landmarks_per_face = landmarks_per_face;
-  result->Reserve(infer_result.shape[1]);
  if (infer_result.dtype != FDDataType::FP32) {
    FDERROR << "Only support post process with float32 data." << std::endl;
    return false;
  }
+
+  result->Clear();
+  // must be setup landmarks_per_face before reserve
+  result->landmarks_per_face = landmarks_per_face;
+  result->Reserve(infer_result.shape[1]);
+
  float* data = static_cast<float*>(infer_result.Data());
  for (size_t i = 0; i < infer_result.shape[1]; ++i) {
    float* reg_cls_ptr = data + (i * infer_result.shape[2]);
--- a/csrcs/fastdeploy/vision/linzaer/ultraface.cc
+++ b/csrcs/fastdeploy/vision/linzaer/ultraface.cc
@@ -106,11 +106,6 @@ bool UltraFace::Postprocess(
  FDTensor& boxes_tensor = infer_result.at(1);   // (1,4420,4)
  FDASSERT((scores_tensor.shape[0] == 1), "Only support batch =1 now.");
  FDASSERT((boxes_tensor.shape[0] == 1), "Only support batch =1 now.");
-
-  result->Clear();
-  // must be setup landmarks_per_face before reserve.
-  // ultraface detector does not detect landmarks by default.
-  result->landmarks_per_face = 0;
  if (scores_tensor.dtype != FDDataType::FP32) {
    FDERROR << "Only support post process with float32 data." << std::endl;
    return false;
@@ -120,6 +115,12 @@ bool UltraFace::Postprocess(
    return false;
  }

+  result->Clear();
+  // must be setup landmarks_per_face before reserve.
+  // ultraface detector does not detect landmarks by default.
+  result->landmarks_per_face = 0;
+  result->Reserve(boxes_tensor.shape[1]);
+
  float* scores_ptr = static_cast<float*>(scores_tensor.Data());
  float* boxes_ptr = static_cast<float*>(boxes_tensor.Data());
  const size_t num_bboxes = boxes_tensor.shape[1];  // e.g 4420
--- a/csrcs/fastdeploy/vision/vision_pybind.cc
+++ b/csrcs/fastdeploy/vision/vision_pybind.cc
@@ -26,6 +26,7 @@ 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);
 #ifdef ENABLE_VISION_VISUALIZE
 void BindVisualize(pybind11::module& m);
 #endif
@@ -71,6 +72,7 @@ void BindVision(pybind11::module& m) {
  BindDeepCam(m);
  BindRangiLyu(m);
  BindLinzaer(m);
+  BindBiubug6(m);
 #ifdef ENABLE_VISION_VISUALIZE
  BindVisualize(m);
 #endif
--- a/examples/vision/biubug6_retinaface.cc
+++ b/examples/vision/biubug6_retinaface.cc
@@ -0,0 +1,55 @@
+// 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/vision.h"
+
+int main() {
+  namespace vis = fastdeploy::vision;
+
+  std::string model_file =
+      "../resources/models/Pytorch_RetinaFace_resnet50-720-1080.onnx";
+  std::string img_path = "../resources/images/test_face_det.jpg";
+  std::string vis_path =
+      "../resources/outputs/biubug6_retinaface_vis_result.jpg";
+
+  auto model = vis::biubug6::RetinaFace(model_file);
+  model.size = {1080, 720};  // (width, height)
+  if (!model.Initialized()) {
+    std::cerr << "Init Failed! Model: " << model_file << std::endl;
+    return -1;
+  } else {
+    std::cout << "Init Done! Model:" << model_file << std::endl;
+  }
+  model.EnableDebug();
+
+  cv::Mat im = cv::imread(img_path);
+  cv::Mat vis_im = im.clone();
+
+  vis::FaceDetectionResult res;
+  if (!model.Predict(&im, &res, 0.3f, 0.3f)) {
+    std::cerr << "Prediction Failed." << std::endl;
+    return -1;
+  } else {
+    std::cout << "Prediction Done!" << std::endl;
+  }
+
+  // 输出预测框结果
+  std::cout << res.Str() << std::endl;
+
+  // 可视化预测结果
+  vis::Visualize::VisFaceDetection(&vis_im, res, 2, 0.3f);
+  cv::imwrite(vis_path, vis_im);
+  std::cout << "Detect Done! Saved: " << vis_path << std::endl;
+  return 0;
+}
--- a/fastdeploy/vision/init.py
+++ b/fastdeploy/vision/init.py
@@ -25,3 +25,4 @@ from . import wongkinyiu
 from . import deepcam
 from . import rangilyu
 from . import linzaer
+from . import biubug6
--- a/fastdeploy/vision/biubug6/init.py
+++ b/fastdeploy/vision/biubug6/init.py
@@ -0,0 +1,98 @@
+# 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 fastdeploy_main as C
+
+
+class RetinaFace(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(RetinaFace, self).__init__(runtime_option)
+
+        self._model = C.vision.biubug6.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."
+
+    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UltraFace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, 480]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 variance(self):
+        return self._model.variance
+
+    @property
+    def downsample_strides(self):
+        return self._model.downsample_strides
+
+    @property
+    def min_sizes(self):
+        return self._model.min_sizes
+
+    @property
+    def landmarks_per_face(self):
+        return self._model.landmarks_per_face
+
+    @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
+
+    @variance.setter
+    def variance(self, value):
+        assert isinstance(v, [list, tuple]),\
+            "The value to set `variance` must be type of tuple or list."
+        assert len(value) == 2,\
+            "The value to set `variance` must contatins 2 elements".format(
+            len(value))
+        self._model.variance = 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
+
+    @min_sizes.setter
+    def min_sizes(self, value):
+        assert isinstance(
+            value, list), "The value to set `min_sizes` must be type of list."
+        self._model.min_sizes = 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
--- a/model_zoo/vision/retinaface/README.md
+++ b/model_zoo/vision/retinaface/README.md
@@ -0,0 +1,76 @@
+# RetinaFace部署示例
+
+当前支持模型版本为：[RetinaFace CommitID:b984b4b](https://github.com/biubug6/Pytorch_Retinaface/commit/b984b4b)
+
+本文档说明如何进行[RetinaFace](https://github.com/biubug6/Pytorch_Retinaface)的快速部署推理。本目录结构如下
+
+```
+.
+├── cpp                     # C++ 代码目录
+│   ├── CMakeLists.txt      # C++ 代码编译CMakeLists文件
+│   ├── README.md           # C++ 代码编译部署文档
+│   └── retinaface.cc       # C++ 示例代码
+├── api.md                  # API 说明文档
+├── README.md               # RetinaFace 部署文档
+└── retinaface.py           # Python示例代码
+```
+
+## 安装FastDeploy
+
+使用如下命令安装FastDeploy，注意到此处安装的是`vision-cpu`，也可根据需求安装`vision-gpu`
+```bash
+# 安装fastdeploy-python工具
+pip install fastdeploy-python
+
+# 安装vision-cpu模块
+fastdeploy install vision-cpu
+```
+
+## Python部署
+
+执行如下代码即会自动下载RetinaFace模型和测试图片
+```bash
+python retinaface.py
+```
+
+## 手动获取ONNX模型文件
+自动下载的模型文件是我们事先转换好的，如果您需要从RetinaFace官方repo导出ONNX，请参考以下步骤。  
+
+* 下载官方仓库并
+```bash
+git clone https://github.com/biubug6/Pytorch_Retinaface.git
+```
+* 下载预训练权重并放在weights文件夹
+```text
+./weights/
+      mobilenet0.25_Final.pth
+      mobilenetV1X0.25_pretrain.tar
+      Resnet50_Final.pth
+```
+* 运行convert_to_onnx.py导出ONNX模型文件
+```bash
+PYTHONPATH=. python convert_to_onnx.py --trained_model ./weights/mobilenet0.25_Final.pth --network mobile0.25 --long_side 640 --cpu
+PYTHONPATH=. python convert_to_onnx.py --trained_model ./weights/Resnet50_Final.pth --network resnet50 --long_side 640 --cpu
+```
+注意：需要先对convert_to_onnx.py脚本中的--long_side参数增加类型约束，type=int.
+* 使用onnxsim对模型进行简化
+```bash
+onnxsim FaceDetector.onnx Pytorch_RetinaFace_mobile0.25-640-640.onnx  # mobilenet
+onnxsim FaceDetector.onnx Pytorch_RetinaFace_resnet50-640-640.onnx  # resnet50
+```
+
+
+执行完成后会将可视化结果保存在本地`vis_result.jpg`，同时输出检测结果如下
+```
+FaceDetectionResult: [xmin, ymin, xmax, ymax, score, (x, y) x 5]
+403.339783,254.192413, 490.002747, 351.931213, 0.999427, (425.657257,293.820740), (467.249451,293.667267), (446.830078,315.016388), (428.903381,326.129425), (465.764648,325.837341)
+296.834564,181.992035, 384.516876, 277.461243, 0.999194, (313.605164,224.800110), (352.888977,219.088043), (333.530182,239.872787), (325.395203,255.463852), (358.417175,250.529892)
+742.206238,263.547424, 840.871765, 366.171387, 0.999068, (762.715759,308.939880), (809.019653,304.544830), (786.174194,329.286163), (771.952271,341.376038), (812.717529,337.528839)
+545.351685,228.015930, 635.423584, 335.458649, 0.998681, (559.295654,269.971619), (598.439758,273.823608), (567.496643,292.894348), (558.160034,306.637238), (592.175781,309.493591)
+180.078125,241.787888, 257.213135, 320.321777, 0.998342, (203.702591,272.032715), (237.497726,271.356445), (222.380402,288.225708), (208.015259,301.360352), (233.943451,300.801636)
+```
+
+## 其它文档
+
+- [C++部署](./cpp/README.md)
+- [RetinaFace API文档](./api.md)
--- a/model_zoo/vision/retinaface/api.md
+++ b/model_zoo/vision/retinaface/api.md
@@ -0,0 +1,71 @@
+# RetinaFace API说明
+
+## Python API
+
+### RetinaFace类
+```
+fastdeploy.vision.biubug6.RetinaFace(model_file, params_file=None, runtime_option=None, model_format=fd.Frontend.ONNX)
+```
+RetinaFace模型加载和初始化，当model_format为`fd.Frontend.ONNX`时，只需提供model_file，如`Pytorch_RetinaFace_mobile0.25-640-640.onnx`；当model_format为`fd.Frontend.PADDLE`时，则需同时提供model_file和params_file。
+
+**参数**
+
+> * **model_file**(str): 模型文件路径
+> * **params_file**(str): 参数文件路径
+> * **runtime_option**(RuntimeOption): 后端推理配置，默认为None，即采用默认配置
+> * **model_format**(Frontend): 模型格式
+
+#### predict函数
+> ```
+> RetinaFace.predict(image_data, conf_threshold=0.7, nms_iou_threshold=0.3)
+> ```
+> 模型预测结口，输入图像直接输出检测结果。
+>
+> **参数**
+>
+> > * **image_data**(np.ndarray): 输入数据，注意需为HWC，BGR格式
+> > * **conf_threshold**(float): 检测框置信度过滤阈值
+> > * **nms_iou_threshold**(float): NMS处理过程中iou阈值
+
+示例代码参考[retinaface.py](./retinaface.py)
+
+
+## C++ API
+
+### RetinaFace 类
+```
+fastdeploy::vision::biubug6::RetinaFace(
+        const string& model_file,
+        const string& params_file = "",
+        const RuntimeOption& runtime_option = RuntimeOption(),
+        const Frontend& model_format = Frontend::ONNX)
+```
+RetinaFace模型加载和初始化，当model_format为`Frontend::ONNX`时，只需提供model_file，如`Pytorch_RetinaFace_mobile0.25-640-640.onnx`；当model_format为`Frontend::PADDLE`时，则需同时提供model_file和params_file。
+
+**参数**
+
+> * **model_file**(str): 模型文件路径
+> * **params_file**(str): 参数文件路径
+> * **runtime_option**(RuntimeOption): 后端推理配置，默认为None，即采用默认配置
+> * **model_format**(Frontend): 模型格式
+
+#### Predict函数
+> ```
+> RetinaFace::Predict(cv::Mat* im, FaceDetectionResult* result,
+>                     float conf_threshold = 0.7,
+>                     float nms_iou_threshold = 0.3)
+> ```
+> 模型预测接口，输入图像直接输出检测结果。
+>
+> **参数**
+>
+> > * **im**: 输入图像，注意需为HWC，BGR格式
+> > * **result**: 检测结果，包括检测框，各个框的置信度
+> > * **conf_threshold**: 检测框置信度过滤阈值
+> > * **nms_iou_threshold**: NMS处理过程中iou阈值
+
+示例代码参考[cpp/retinaface.cc](cpp/retinaface.cc)
+
+## 其它API使用
+
+- [模型部署RuntimeOption配置](../../../docs/api/runtime_option.md)
--- a/model_zoo/vision/retinaface/cpp/CMakeLists.txt
+++ b/model_zoo/vision/retinaface/cpp/CMakeLists.txt
@@ -0,0 +1,17 @@
+PROJECT(retinaface_demo C CXX)
+CMAKE_MINIMUM_REQUIRED (VERSION 3.16)
+
+# 在低版本ABI环境中，通过如下代码进行兼容性编译
+# add_definitions(-D_GLIBCXX_USE_CXX11_ABI=0)
+
+# 指定下载解压后的fastdeploy库路径
+set(FASTDEPLOY_INSTALL_DIR ${PROJECT_SOURCE_DIR}/fastdeploy-linux-x64-0.3.0/)
+
+include(${FASTDEPLOY_INSTALL_DIR}/FastDeploy.cmake)
+
+# 添加FastDeploy依赖头文件
+include_directories(${FASTDEPLOY_INCS})
+
+add_executable(retinaface_demo ${PROJECT_SOURCE_DIR}/retinaface.cc)
+# 添加FastDeploy库依赖
+target_link_libraries(retinaface_demo ${FASTDEPLOY_LIBS})
--- a/model_zoo/vision/retinaface/cpp/README.md
+++ b/model_zoo/vision/retinaface/cpp/README.md
@@ -0,0 +1,61 @@
+# 编译RetinaFace示例
+
+当前支持模型版本为：[RetinaFace CommitID:b984b4b](https://github.com/biubug6/Pytorch_Retinaface/commit/b984b4b)
+
+## 下载和解压预测库
+```bash
+wget https://bj.bcebos.com/paddle2onnx/fastdeploy/fastdeploy-linux-x64-0.0.3.tgz
+tar xvf fastdeploy-linux-x64-0.0.3.tgz
+```
+
+## 编译示例代码
+```bash
+mkdir build & cd build
+cmake ..
+make -j
+```
+
+## 下载模型和图片
+wget https://github.com/DefTruth/Pytorch_Retinaface/releases/download/v0.1/Pytorch_RetinaFace_mobile0.25-640-640.onnx  
+wget https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB/raw/master/imgs/3.jpg
+
+## 手动获取ONNX模型文件
+自动下载的模型文件是我们事先转换好的，如果您需要从RetinaFace官方repo导出ONNX，请参考以下步骤。  
+
+* 下载官方仓库并
+```bash
+git clone https://github.com/biubug6/Pytorch_Retinaface.git
+```
+* 下载预训练权重并放在weights文件夹
+```text
+./weights/
+      mobilenet0.25_Final.pth
+      mobilenetV1X0.25_pretrain.tar
+      Resnet50_Final.pth
+```
+* 运行convert_to_onnx.py导出ONNX模型文件
+```bash
+PYTHONPATH=. python convert_to_onnx.py --trained_model ./weights/mobilenet0.25_Final.pth --network mobile0.25 --long_side 640 --cpu
+PYTHONPATH=. python convert_to_onnx.py --trained_model ./weights/Resnet50_Final.pth --network resnet50 --long_side 640 --cpu
+```
+注意：需要先对convert_to_onnx.py脚本中的--long_side参数增加类型约束，type=int.
+* 使用onnxsim对模型进行简化
+```bash
+onnxsim FaceDetector.onnx Pytorch_RetinaFace_mobile0.25-640-640.onnx  # mobilenet
+onnxsim FaceDetector.onnx Pytorch_RetinaFace_resnet50-640-640.onnx  # resnet50
+```
+
+## 执行
+```bash
+./retinaface_demo
+```
+
+执行完后可视化的结果保存在本地`vis_result.jpg`，同时会将检测框输出在终端，如下所示
+```
+FaceDetectionResult: [xmin, ymin, xmax, ymax, score, (x, y) x 5]
+403.339783,254.192413, 490.002747, 351.931213, 0.999427, (425.657257,293.820740), (467.249451,293.667267), (446.830078,315.016388), (428.903381,326.129425), (465.764648,325.837341)
+296.834564,181.992035, 384.516876, 277.461243, 0.999194, (313.605164,224.800110), (352.888977,219.088043), (333.530182,239.872787), (325.395203,255.463852), (358.417175,250.529892)
+742.206238,263.547424, 840.871765, 366.171387, 0.999068, (762.715759,308.939880), (809.019653,304.544830), (786.174194,329.286163), (771.952271,341.376038), (812.717529,337.528839)
+545.351685,228.015930, 635.423584, 335.458649, 0.998681, (559.295654,269.971619), (598.439758,273.823608), (567.496643,292.894348), (558.160034,306.637238), (592.175781,309.493591)
+180.078125,241.787888, 257.213135, 320.321777, 0.998342, (203.702591,272.032715), (237.497726,271.356445), (222.380402,288.225708), (208.015259,301.360352), (233.943451,300.801636)
+```
--- a/model_zoo/vision/retinaface/cpp/retinaface.cc
+++ b/model_zoo/vision/retinaface/cpp/retinaface.cc
@@ -0,0 +1,49 @@
+// 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/vision.h"
+
+int main() {
+  namespace vis = fastdeploy::vision;
+
+  auto model =
+      vis::biubug6::RetinaFace("Pytorch_RetinaFace_mobile0.25-640-640.onnx");
+  if (!model.Initialized()) {
+    std::cerr << "Init Failed! Model: " << model_file << std::endl;
+    return -1;
+  } else {
+    std::cout << "Init Done! Model:" << model_file << std::endl;
+  }
+  model.EnableDebug();
+
+  cv::Mat im = cv::imread("3.jpg");
+  cv::Mat vis_im = im.clone();
+
+  vis::FaceDetectionResult res;
+  if (!model.Predict(&im, &res, 0.7f, 0.3f)) {
+    std::cerr << "Prediction Failed." << std::endl;
+    return -1;
+  } else {
+    std::cout << "Prediction Done!" << std::endl;
+  }
+
+  // 输出预测框结果
+  std::cout << res.Str() << std::endl;
+
+  // 可视化预测结果
+  vis::Visualize::VisFaceDetection(&vis_im, res, 2, 0.3f);
+  cv::imwrite("vis_result.jpg", vis_im);
+  std::cout << "Detect Done! Saved: " << vis_path << std::endl;
+  return 0;
+}
--- a/model_zoo/vision/retinaface/retinaface.py
+++ b/model_zoo/vision/retinaface/retinaface.py
@@ -0,0 +1,24 @@
+import fastdeploy as fd
+import cv2
+
+# 下载模型
+model_url = "https://github.com/DefTruth/Pytorch_Retinaface/releases/download/v0.1/Pytorch_RetinaFace_mobile0.25-640-640.onnx"
+test_img_url = "https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB/raw/master/imgs/3.jpg"
+fd.download(model_url, ".", show_progress=True)
+fd.download(test_img_url, ".", show_progress=True)
+
+# 加载模型
+model = fd.vision.biubug6.RetinaFace(
+    "Pytorch_RetinaFace_mobile0.25-640-640.onnx")
+
+# 预测图片
+im = cv2.imread("3.jpg")
+result = model.predict(im, conf_threshold=0.7, nms_iou_threshold=0.3)
+
+# 可视化结果
+fd.vision.visualize.vis_face_detection(im, result)
+cv2.imwrite("vis_result.jpg", im)
+
+# 输出预测结果
+print(result)
+print(model.runtime_option)
--- a/model_zoo/vision/ultraface/README.md
+++ b/model_zoo/vision/ultraface/README.md
@@ -28,7 +28,7 @@ fastdeploy install vision-cpu

 ## Python部署

-执行如下代码即会自动下载YOLOv5Face模型和测试图片
+执行如下代码即会自动下载UltraFace模型和测试图片
 ```bash
 python ultraface.py
 ```
--- a/model_zoo/vision/yolov5face/api.md
+++ b/model_zoo/vision/yolov5face/api.md
@@ -51,7 +51,7 @@ YOLOv5Face模型加载和初始化，当model_format为`Frontend::ONNX`时，只

 #### Predict函数
 > ```
-> YOLOv5Face::Predict(cv::Mat* im, DetectionResult* result,
+> YOLOv5Face::Predict(cv::Mat* im, FaceDetectionResult* result,
 >                     float conf_threshold = 0.25,
 >                     float nms_iou_threshold = 0.5)
 > ```