Add YOLOv5-cls Model (#335)

* add yolov5cls * fixed bugs * fixed bugs * fixed preprocess bug * add yolov5cls readme * deal with comments * Add YOLOv5Cls Note * add yolov5cls test Co-authored-by: Jason <jiangjiajun@baidu.com>
2025-10-05 08:37:06 +08:00 · 2022-10-12 15:57:26 +08:00
parent 945e197bd1
commit b557dbc2d8
19 changed files with 719 additions and 4 deletions
--- a/examples/vision/classification/paddleclas/python/README.md
+++ b/examples/vision/classification/paddleclas/python/README.md
@@ -55,7 +55,7 @@ PaddleClas模型加载和初始化，其中model_file, params_file为训练模
 > PaddleClasModel.predict(input_image, topk=1)
 > ```
 >
-> 模型预测结口，输入图像直接输出检测结果。
+> 模型预测结口，输入图像直接输出分类topk结果。
 >
 > **参数**
 >
--- a/examples/vision/classification/yolov5cls/README.md
+++ b/examples/vision/classification/yolov5cls/README.md
@@ -0,0 +1,29 @@
+# YOLOv5Cls准备部署模型
+
+- YOLOv5Cls v6.2部署模型实现来自[YOLOv5](https://github.com/ultralytics/yolov5/tree/v6.2),和[基于ImageNet的预训练模型](https://github.com/ultralytics/yolov5/releases/tag/v6.2)
+  - （1）[官方库](https://github.com/ultralytics/yolov5/releases/tag/v6.2)提供的*-cls.pt模型，使用[YOLOv5](https://github.com/ultralytics/yolov5)中的`export.py`导出ONNX文件后，可直接进行部署；
+  - （2）开发者基于自己数据训练的YOLOv5Cls v6.2模型，可使用[YOLOv5](https://github.com/ultralytics/yolov5)中的`export.py`导出ONNX文件后，完成部署。
+
+
+## 下载预训练ONNX模型
+
+为了方便开发者的测试，下面提供了YOLOv5Cls导出的各系列模型，开发者可直接下载使用。（下表中模型的精度来源于源官方库）
+| 模型                                                               | 大小    | 精度(top1)  | 精度(top5)    |
+|:---------------------------------------------------------------- |:----- |:----- |:----- |
+| [YOLOv5n-cls](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5n-cls.onnx) | 9.6MB | 64.6% | 85.4% |
+| [YOLOv5s-cls](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5s-cls.onnx) | 21MB | 71.5% | 90.2% |
+| [YOLOv5m-cls](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5m-cls.onnx) | 50MB | 75.9% | 92.9% |
+| [YOLOv5l-cls](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5l-cls.onnx) | 102MB | 78.0% | 94.0% |
+| [YOLOv5x-cls](https://bj.bcebos.com/paddlehub/fastdeploy/yolov5x-cls.onnx) | 184MB | 79.0% | 94.4% |
+
+
+
+
+## 详细部署文档
+
+- [Python部署](python)
+- [C++部署](cpp)
+
+## 版本说明
+
+- 本版本文档和代码基于[YOLOv5 v6.2](https://github.com/ultralytics/yolov5/tree/v6.2) 编写
--- a/examples/vision/classification/yolov5cls/cpp/CMakeLists.txt
+++ b/examples/vision/classification/yolov5cls/cpp/CMakeLists.txt
@@ -0,0 +1,14 @@
+PROJECT(infer_demo C CXX)
+CMAKE_MINIMUM_REQUIRED (VERSION 3.12)
+
+# 指定下载解压后的fastdeploy库路径
+option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
+
+include(${FASTDEPLOY_INSTALL_DIR}/FastDeploy.cmake)
+
+# 添加FastDeploy依赖头文件
+include_directories(${FASTDEPLOY_INCS})
+
+add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer.cc)
+# 添加FastDeploy库依赖
+target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
--- a/examples/vision/classification/yolov5cls/cpp/README.md
+++ b/examples/vision/classification/yolov5cls/cpp/README.md
@@ -0,0 +1,89 @@
+# YOLOv5Cls C++部署示例
+
+本目录下提供`infer.cc`快速完成YOLOv5Cls在CPU/GPU，以及GPU上通过TensorRT加速部署的示例。
+
+在部署前，需确认以下两个步骤
+
+- 1. 软硬件环境满足要求，参考[FastDeploy环境要求](../../../../../docs/environment.md)  
+- 2. 根据开发环境，下载预编译部署库和samples代码，参考[FastDeploy预编译库](../../../../../docs/quick_start)
+
+以Linux上CPU推理为例，在本目录执行如下命令即可完成编译测试
+
+```bash
+mkdir build
+cd build
+wget https://bj.bcebos.com/fastdeploy/release/cpp/fastdeploy-linux-x64-0.2.1.tgz
+tar xvf fastdeploy-linux-x64-0.2.1.tgz
+cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/fastdeploy-linux-x64-0.2.1
+make -j
+
+#下载官方转换好的yolov5模型文件和测试图片
+wget https://bj.bcebos.com/paddlehub/fastdeploy/yolov5n-cls.onnx
+wget hhttps://gitee.com/paddlepaddle/PaddleClas/raw/release/2.4/deploy/images/ImageNet/ILSVRC2012_val_00000010.jpeg
+
+
+# CPU推理
+./infer_demo yolov5n-cls.onnx 000000014439.jpg 0
+# GPU推理
+./infer_demo yolov5n-cls.onnx 000000014439.jpg 1
+# GPU上TensorRT推理
+./infer_demo yolov5n-cls.onnx 000000014439.jpg 2
+```
+
+运行完成后返回结果如下所示
+```bash
+ClassifyResult(
+label_ids: 265,
+scores: 0.196327,
+)
+```
+
+以上命令只适用于Linux或MacOS, Windows下SDK的使用方式请参考:  
+- [如何在Windows中使用FastDeploy C++ SDK](../../../../../docs/compile/how_to_use_sdk_on_windows.md)
+
+## YOLOv5Cls C++接口
+
+### YOLOv5Cls类
+
+```c++
+fastdeploy::vision::classification::YOLOv5Cls(
+        const string& model_file,
+        const string& params_file = "",
+        const RuntimeOption& runtime_option = RuntimeOption(),
+        const ModelFormat& model_format = ModelFormat::ONNX)
+```
+
+YOLOv5Cls模型加载和初始化，其中model_file为导出的ONNX模型格式。
+
+**参数**
+
+> * **model_file**(str): 模型文件路径
+> * **params_file**(str): 参数文件路径，当模型格式为ONNX时，此参数传入空字符串即可
+> * **runtime_option**(RuntimeOption): 后端推理配置，默认为None，即采用默认配置
+> * **model_format**(ModelFormat): 模型格式，默认为ONNX格式
+
+#### Predict函数
+
+> ```c++
+> YOLOv5Cls::Predict(cv::Mat* im, int topk = 1)
+> ```
+>
+> 模型预测接口，输入图像直接输出输出分类topk结果。
+>
+> **参数**
+>
+> > * **input_image**(np.ndarray): 输入数据，注意需为HWC，BGR格式
+> > * **topk**(int):返回预测概率最高的topk个分类结果，默认为1
+
+
+> **返回**
+>
+> > 返回`fastdeploy.vision.ClassifyResult`结构体，结构体说明参考文档[视觉模型预测结果](../../../../../docs/api/vision_results/)
+
+
+## 其它文档
+
+- [YOLOv5Cls 模型介绍](..)
+- [YOLOv5Cls Python部署](../python)
+- [模型预测结果说明](../../../../../docs/api/vision_results/)
+- [如何切换模型推理后端引擎](../../../../../docs/runtime/how_to_change_backend.md)
--- a/examples/vision/classification/yolov5cls/cpp/infer.cc
+++ b/examples/vision/classification/yolov5cls/cpp/infer.cc
@@ -0,0 +1,104 @@
+// 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"
+#ifdef WIN32
+const char sep = '\\';
+#else
+const char sep = '/';
+#endif
+
+void CpuInfer(const std::string& model_file, const std::string& image_file) {
+  auto model = fastdeploy::vision::classification::YOLOv5Cls(model_file);
+  if (!model.Initialized()) {
+    std::cerr << "Failed to initialize." << std::endl;
+    return;
+  }
+
+  auto im = cv::imread(image_file);
+  auto im_bak = im.clone();
+
+  fastdeploy::vision::ClassifyResult res;
+  if (!model.Predict(&im, &res)) {
+    std::cerr << "Failed to predict." << std::endl;
+    return;
+  }
+  // print res
+  std::cout << res.Str() << std::endl;
+}
+
+void GpuInfer(const std::string& model_file, const std::string& image_file) {
+  auto option = fastdeploy::RuntimeOption();
+  option.UseGpu();
+  auto model = fastdeploy::vision::classification::YOLOv5Cls(model_file, "", option);
+  if (!model.Initialized()) {
+    std::cerr << "Failed to initialize." << std::endl;
+    return;
+  }
+
+  auto im = cv::imread(image_file);
+  auto im_bak = im.clone();
+
+  fastdeploy::vision::ClassifyResult res;
+  if (!model.Predict(&im, &res)) {
+    std::cerr << "Failed to predict." << std::endl;
+    return;
+  }
+  // print res
+  std::cout << res.Str() << std::endl;
+}
+
+void TrtInfer(const std::string& model_file, const std::string& image_file) {
+  auto option = fastdeploy::RuntimeOption();
+  option.UseGpu();
+  option.UseTrtBackend();
+  option.SetTrtInputShape("images", {1, 3, 224, 224});
+  auto model = fastdeploy::vision::classification::YOLOv5Cls(model_file, "", option);
+  if (!model.Initialized()) {
+    std::cerr << "Failed to initialize." << std::endl;
+    return;
+  }
+
+  auto im = cv::imread(image_file);
+  auto im_bak = im.clone();
+
+  fastdeploy::vision::ClassifyResult res;
+  if (!model.Predict(&im, &res)) {
+    std::cerr << "Failed to predict." << std::endl;
+    return;
+  }
+  // print res
+  std::cout << res.Str() << std::endl;
+}
+
+int main(int argc, char* argv[]) {
+  if (argc < 4) {
+    std::cout << "Usage: infer_demo path/to/model path/to/image run_option, "
+                 "e.g ./infer_model ./yolov5n-cls.onnx ./test.jpeg 0"
+              << std::endl;
+    std::cout << "The data type of run_option is int, 0: run with cpu; 1: run "
+                 "with gpu; 2: run with gpu and use tensorrt backend."
+              << std::endl;
+    return -1;
+  }
+
+  if (std::atoi(argv[3]) == 0) {
+    CpuInfer(argv[1], argv[2]);
+  } else if (std::atoi(argv[3]) == 1) {
+    GpuInfer(argv[1], argv[2]);
+  } else if (std::atoi(argv[3]) == 2) {
+    TrtInfer(argv[1], argv[2]);
+  }
+  return 0;
+}
--- a/examples/vision/classification/yolov5cls/python/README.md
+++ b/examples/vision/classification/yolov5cls/python/README.md
@@ -0,0 +1,73 @@
+# YOLOv5Cls Python部署示例
+
+在部署前，需确认以下两个步骤
+
+- 1. 软硬件环境满足要求，参考[FastDeploy环境要求](../../../../../docs/environment.md)  
+- 2. FastDeploy Python whl包安装，参考[FastDeploy Python安装](../../../../../docs/quick_start)
+
+本目录下提供`infer.py`快速完成YOLOv5Cls在CPU/GPU，以及GPU上通过TensorRT加速部署的示例。执行如下脚本即可完成
+
+```bash
+#下载部署示例代码
+git clone https://github.com/PaddlePaddle/FastDeploy.git
+cd examples/vision/classification/yolov5cls/python/
+
+#下载yolov5cls模型文件和测试图片
+wget https://bj.bcebos.com/paddlehub/fastdeploy/yolov5n-cls.onnx
+wget https://gitee.com/paddlepaddle/PaddleClas/raw/release/2.4/deploy/images/ImageNet/ILSVRC2012_val_00000010.jpeg
+
+# CPU推理
+python infer.py --model yolov5n-cls.onnx --image ILSVRC2012_val_00000010.jpeg --device cpu --topk 1
+# GPU推理
+python infer.py --model yolov5n-cls.onnx --image ILSVRC2012_val_00000010.jpeg --device gpu --topk 1
+# GPU上使用TensorRT推理
+python infer.py --model yolov5n-cls.onnx --image ILSVRC2012_val_00000010.jpeg --device gpu --use_trt True
+```
+
+运行完成后返回结果如下所示
+```bash
+ClassifyResult(
+label_ids: 265,
+scores: 0.196327,
+)
+```
+
+## YOLOv5Cls Python接口
+
+```python
+fastdeploy.vision.classification.YOLOv5Cls(model_file, params_file=None, runtime_option=None, model_format=ModelFormat.ONNX)
+```
+
+YOLOv5Cls模型加载和初始化，其中model_file为导出的ONNX模型格式
+
+**参数**
+
+> * **model_file**(str): 模型文件路径
+> * **params_file**(str): 参数文件路径，当模型格式为ONNX格式时，此参数无需设定
+> * **runtime_option**(RuntimeOption): 后端推理配置，默认为None，即采用默认配置
+> * **model_format**(ModelFormat): 模型格式，默认为ONNX
+
+### predict函数
+
+> ```python
+> YOLOv5Cls.predict(image_data, topk=1)
+> ```
+>
+> 模型预测结口，输入图像直接输出分类topk结果。
+>
+> **参数**
+>
+> > * **input_image**(np.ndarray): 输入数据，注意需为HWC，BGR格式
+> > * **topk**(int):返回预测概率最高的topk个分类结果，默认为1
+
+> **返回**
+>
+> > 返回`fastdeploy.vision.ClassifyResult`结构体，结构体说明参考文档[视觉模型预测结果](../../../../../docs/api/vision_results/)
+
+
+## 其它文档
+
+- [YOLOv5Cls 模型介绍](..)
+- [YOLOv5Cls C++部署](../cpp)
+- [模型预测结果说明](../../../../../docs/api/vision_results/)
+- [如何切换模型推理后端引擎](../../../../../docs/runtime/how_to_change_backend.md)
--- a/examples/vision/classification/yolov5cls/python/infer.py
+++ b/examples/vision/classification/yolov5cls/python/infer.py
@@ -0,0 +1,51 @@
+import fastdeploy as fd
+import cv2
+import os
+
+
+def parse_arguments():
+    import argparse
+    import ast
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--model", required=True, help="Path of YOLOv5Cls model.")
+    parser.add_argument(
+        "--image", type=str, required=True, help="Path of test image file.")
+    parser.add_argument(
+        "--topk", type=int, default=1, help="Return topk results.")
+    parser.add_argument(
+        "--device",
+        type=str,
+        default='cpu',
+        help="Type of inference device, support 'cpu' or 'gpu'.")
+    parser.add_argument(
+        "--use_trt",
+        type=ast.literal_eval,
+        default=False,
+        help="Wether to use tensorrt.")
+    return parser.parse_args()
+
+
+def build_option(args):
+    option = fd.RuntimeOption()
+
+    if args.device.lower() == "gpu":
+        option.use_gpu()
+
+    if args.use_trt:
+        option.use_trt_backend()
+        option.set_trt_input_shape("images", [1, 3, 224, 224])
+    return option
+
+
+args = parse_arguments()
+
+# 配置runtime，加载模型
+runtime_option = build_option(args)
+model = fd.vision.classification.YOLOv5Cls(
+    args.model, runtime_option=runtime_option)
+
+# 预测图片分类结果
+im = cv2.imread(args.image)
+result = model.predict(im.copy(), args.topk)
+print(result)