[docs][win] add windows c++ sdk demo to examples (#136)

* [docs] format docs with markdown with language tags * [docs][win] add windows c++ sdk demo * [docs][win] add windows c++ sdk demo to examples * [docs][api] update runtime_option docs
2025-10-05 08:37:06 +08:00 · 2022-08-22 10:53:00 +08:00
parent f7e4b8490f
commit beddcba900
54 changed files with 278 additions and 164 deletions
--- a/README.md
+++ b/README.md
@@ -19,7 +19,7 @@
 ## 近期更新

 - 🔥 **2022.8.18：发布FastDeploy [release/v0.2.0](https://github.com/PaddlePaddle/FastDeploy/releases/tag/release%2F0.2.0)** <br>
-    - **服务端全新升级：一套SDK，覆盖全量模型**   
+    - **服务端全新升级：一套SDK，覆盖全量模型**  
        - 发布基于x86 CPU、NVIDIA GPU的易用、高性能推理引擎SDK，推理速度大幅提升
        - 支持ONNXRuntime、Paddle Inference、TensorRT推理引擎
        - 支持YOLOv7、YOLOv6、YOLOv5、PP-YOLOE等目标检测最优模型及[Demo示例](examples/vision/detection/)
@@ -51,7 +51,7 @@

 <div id="fastdeploy-quick-start"></div>

-### 1.1 快速安装 FastDeploy Python/C++ 库 
+### 1.1 快速安装 FastDeploy Python/C++ 库

 #### 环境依赖

@@ -63,11 +63,11 @@
 - python 3.6\~3.9(Windows 10 3.8\~3.9)

 #### 安装 CPU Python 版本
-```
+```bash
 pip install numpy opencv-python fastdeploy-python -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html
 ```
 #### 安装 GPU Python 版本
-```
+```bash
 pip install numpy opencv-python fastdeploy-gpu-python -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html
 ```
 #### 安装 C++ 版本
@@ -93,8 +93,8 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000
 import cv2
 import fastdeploy.vision as vision

-model = vision.detection.PPYOLOE("ppyoloe_crn_l_300e_coco/model.pdmodel", 
-                                 "ppyoloe_crn_l_300e_coco/model.pdiparams", 
+model = vision.detection.PPYOLOE("ppyoloe_crn_l_300e_coco/model.pdmodel",
+                                 "ppyoloe_crn_l_300e_coco/model.pdiparams",
                                 "ppyoloe_crn_l_300e_coco/nfer_cfg.yml")
 im = cv2.imread("000000014439.jpg")
 result = model.predict(im.copy())
@@ -114,8 +114,8 @@ cv2.imwrite("vis_image.jpg", vis_im)

 int main(int argc, char* argv[]) {
  namespace vision = fastdeploy::vision;
-  auto model = vision::detection::PPYOLOE("ppyoloe_crn_l_300e_coco/model.pdmodel", 
-                                          "ppyoloe_crn_l_300e_coco/model.pdiparams", 
+  auto model = vision::detection::PPYOLOE("ppyoloe_crn_l_300e_coco/model.pdmodel",
+                                          "ppyoloe_crn_l_300e_coco/model.pdiparams",
                                          "ppyoloe_crn_l_300e_coco/infer_cfg.yml");
  auto im = cv::imread("000000014439.jpg");

@@ -128,7 +128,7 @@ int main(int argc, char* argv[]) {
 ```

 更多部署案例请参考[视觉模型部署示例](examples/vision) .
-  
+
 ## 2. 服务端模型列表 🔥🔥🔥

 <div id="fastdeploy-server-models"></div>
--- a/docs/api/runtime_option.md
+++ b/docs/api/runtime_option.md
@@ -15,13 +15,13 @@ FastDeploy产品中的Runtime包含多个推理后端，其各关系如下所示
 |   GPU       |   支持       | 支持       | 支持    | 支持   |

 在各模型的，均通过`RuntimeOption`来配置推理的后端，以及推理时的参数，例如在python中，加载模型后可通过如下代码打印推理配置
-```
-model = fastdeploy.vision.ultralytics.YOLOv5("yolov5s.onnx")
+```python
+model = fastdeploy.vision.detection.YOLOv5("yolov5s.onnx")
 print(model.runtime_option)
 ```
 可看下如下输出

-```
+```python
 RuntimeOption(
  backend : Backend.ORT                # 推理后端ONNXRuntime
  cpu_thread_num : 8                   # CPU线程数（仅当使用CPU推理时有效）
@@ -70,7 +70,7 @@ RuntimeOption(
 > * **trt_max_shape**(dict[str : list[int]]): 当模型为动态shape，且实际推理时输入shape也会变化，通过此参数配置输入的最大shape
 > * **trt_max_batch_size**(int): TensorRT推理时的最大batch数

-```
+```python
 import fastdeploy as fd

 option = fd.RuntimeOption()
@@ -81,10 +81,11 @@ option.trt_min_shape = {"x": [1, 3, 224, 224]}
 option.trt_opt_shape = {"x": [4, 3, 224, 224]}
 option.trt_max_shape = {"x": [8, 3, 224, 224]}

-model = fd.vision.ppcls.Model("resnet50/inference.pdmodel",
-                              "resnet50/inference.pdiparams",
-                              "resnet50/inference_cls.yaml",
-                              runtime_option=option)
+model = fd.vision.classification.PaddleClasModel(
+    "resnet50/inference.pdmodel",
+    "resnet50/inference.pdiparams",
+    "resnet50/inference_cls.yaml",
+    runtime_option=option)
 ```

 ## C++ 使用
@@ -112,7 +113,7 @@ model = fd.vision.ppcls.Model("resnet50/inference.pdmodel",
 > * **trt_max_shape**(map<string, vector<int>>): 当模型为动态shape，且实际推理时输入shape也会变化，通过此参数配置输入的最大shape
 > * **trt_max_batch_size**(int): TensorRT推理时的最大batch数

-```
+```c++
 #include "fastdeploy/vision.h"

 int main() {
@@ -121,11 +122,11 @@ int main() {
  option.trt_opt_shape["x"] = {4, 3, 224, 224};
  option.trt_max_shape["x"] = {8, 3, 224, 224};

-  auto model = fastdeploy::vision::ppcls.Model(
+  auto model = fastdeploy::vision::classification::PaddleClasModel(
                           "resnet50/inference.pdmodel",
                           "resnet50/inference.pdiparams",
                           "resnet50/inference_cls.yaml",
-                           option);
+                            option);
  return 0;
 }
 ```
--- a/docs/api/vision_results/classification_result.md
+++ b/docs/api/vision_results/classification_result.md
@@ -6,7 +6,7 @@ ClassifyResult代码定义在`csrcs/fastdeploy/vision/common/result.h`中，用

 `fastdeploy::vision::ClassifyResult`

-```
+```c++
 struct ClassifyResult {
  std::vector<int32_t> label_ids;
  std::vector<float> scores;
--- a/docs/api/vision_results/detection_result.md
+++ b/docs/api/vision_results/detection_result.md
@@ -6,7 +6,7 @@ DetectionResult代码定义在`csrcs/fastdeploy/vision/common/result.h`中，用

 `fastdeploy::vision::DetectionResult`

-```
+```c++
 struct DetectionResult {
  std::vector<std::array<float, 4>> boxes;
  std::vector<float> scores;
--- a/docs/api/vision_results/face_detection_result.md
+++ b/docs/api/vision_results/face_detection_result.md
@@ -6,7 +6,7 @@ FaceDetectionResult 代码定义在`csrcs/fastdeploy/vision/common/result.h`中

 `fastdeploy::vision::FaceDetectionResult`

-```
+```c++
 struct FaceDetectionResult {
  std::vector<std::array<float, 4>> boxes;
  std::vector<std::array<float, 2>> landmarks;
@@ -32,4 +32,3 @@ struct FaceDetectionResult {
 - **scores**(list of float): 成员变量，表示单张图片检测出来的所有目标置信度
 - **landmarks**(list of list(float)): 成员变量，表示单张图片检测出来的所有人脸的关键点
 - **landmarks_per_face**(int): 成员变量，表示每个人脸框中的关键点的数量。
-
--- a/docs/api/vision_results/face_recognition_result.md
+++ b/docs/api/vision_results/face_recognition_result.md
@@ -5,7 +5,7 @@ FaceRecognitionResult 代码定义在`csrcs/fastdeploy/vision/common/result.h`

 `fastdeploy::vision::FaceRecognitionResult`

-```
+```c++
 struct FaceRecognitionResult {
  std::vector<float> embedding;
  void Clear();
--- a/docs/api/vision_results/matting_result.md
+++ b/docs/api/vision_results/matting_result.md
@@ -6,7 +6,7 @@ MattingResult 代码定义在`csrcs/fastdeploy/vision/common/result.h`中，用

 `fastdeploy::vision::MattingResult`

-```
+```c++
 struct MattingResult {
  std::vector<float> alpha;
  std::vector<float> foreground;
--- a/docs/api/vision_results/segmentation_result.md
+++ b/docs/api/vision_results/segmentation_result.md
@@ -6,7 +6,7 @@ SegmentationResult代码定义在`csrcs/fastdeploy/vision/common/result.h`中，

 `fastdeploy::vision::DetectionResult`

-```
+```c++
 struct DetectionResult {
  std::vector<uint8_t> label_map;
  std::vector<float> score_map;
--- a/docs/compile/how_to_build_linux_and_mac.md
+++ b/docs/compile/how_to_build_linux_and_mac.md
@@ -9,7 +9,7 @@
 - TensorRT >= 8.4 (当ENABLE_TRT_BACKEND=ON)

 ## 编译C++
-```
+```bash
 git clone https://github.com/PaddlePaddle/FastDeploy.git
 cd FastDeploy
 git checkout develop
@@ -23,7 +23,7 @@ make install
 编译后的预测库即在当前目录下的`fastdeploy-0.0.3`

 ## 编译Python安装包
-```
+```bash
 git clone https://github.com/PaddlePaddle/FastDeploy.git
 cd FastDeploy
 git checkout develop
--- a/docs/compile/how_to_build_windows.md
+++ b/docs/compile/how_to_build_windows.md
@@ -3,13 +3,13 @@
 ## 环境依赖

 - cmake >= 3.12
- g++ >= 8.2
+- Visual Studio 16 2019
 - cuda >= 11.2 (当WITH_GPU=ON)
 - cudnn >= 11.2 (当WITH_GPU=ON)
 - TensorRT >= 8.4 (当ENABLE_TRT_BACKEND=ON)

 ## 获取代码
-```
+```bat
 git clone https://github.com/PaddlePaddle/FastDeploy.git
 cd FastDeploy
 git checkout develop
@@ -19,10 +19,10 @@ git checkout develop

 Windows菜单打开`x64 Native Tools Command Prompt for VS 2019`命令工具，其中`CMAKE_INSTALL_PREFIX`用于指定编译后生成的SDK路径

-```
+```bat
 mkdir build
 cd build
-cmake -G "Visual Studio 16 2019" -A x64 -DCMAKE_INSTALL_PREFIX=D:\Paddle\FastDeploy\build\fastdeploy-win-x64-0.2.0 -DENABLE_ORT_BACKEND=ON -DENABLE_VISION=ON .. 
+cmake -G "Visual Studio 16 2019" -A x64 -DCMAKE_INSTALL_PREFIX=D:\Paddle\FastDeploy\build\fastdeploy-win-x64-0.2.0 -DENABLE_ORT_BACKEND=ON -DENABLE_VISION=ON ..
 msbuild fastdeploy.sln /m /p:Configuration=Release /p:Platform=x64
 msbuild INSTALL.vcxproj /m /p:Configuration=Release /p:Platform=x64
 ```
@@ -31,7 +31,7 @@ msbuild INSTALL.vcxproj /m /p:Configuration=Release /p:Platform=x64
 ## 编译Python Wheel包

 Python编译时，通过环境变量获取编译选项
-```
+```bat
 git clone https://github.com/PaddlePaddle/FastDeploy.git
 cd FastDeploy
 git checkout develop
--- a/docs/compile/how_to_use_sdk_on_windows.md
+++ b/docs/compile/how_to_use_sdk_on_windows.md
@@ -0,0 +1,63 @@
+# 在 Windows 使用 FastDeploy C++ SDK
+
+在 Windows 下使用 FastDeploy C++ SDK 与在 Linux 下使用稍有不同。以下以 PPYOLOE 为例进行演示在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/the%20software%20and%20hardware%20requirements.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)
+
+## 环境依赖
+
+- cmake >= 3.12
+- Visual Studio 16 2019
+- cuda >= 11.2 (当WITH_GPU=ON)
+- cudnn >= 11.2 (当WITH_GPU=ON)
+- TensorRT >= 8.4 (当ENABLE_TRT_BACKEND=ON)
+
+## 下载 FastDeploy Windows 10 C++ SDK
+可以从以下链接下载编译好的 FastDeploy Windows 10 C++ SDK，SDK中包含了examples代码。
+```text
+https://bj.bcebos.com/fastdeploy/release/cpp/fastdeploy-win-x64-gpu-0.2.0.zip
+```
+## 准备模型文件和测试图片  
+可以从以下链接下载模型文件和测试图片，并解压缩
+```text
+https://bj.bcebos.com/paddlehub/fastdeploy/ppyoloe_crn_l_300e_coco.tgz # (下载后解压缩)
+https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/000000014439.jpg
+```
+
+## 在 Windows 上编译 PPYOLOE
+Windows菜单打开`x64 Native Tools Command Prompt for VS 2019`命令工具，cd到ppyoloe的demo路径  
+```bat  
+cd fastdeploy-win-x64-gpu-0.2.0\examples\vision\detection\paddledetection\cpp
+```
+```bat
+mkdir build && cd build
+cmake .. -G "Visual Studio 16 2019" -A x64 -DFASTDEPLOY_INSTALL_DIR=%cd%\..\..\..\..\..\..\..\fastdeploy-win-x64-gpu-0.2.0 -DCUDA_DIRECTORY="C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v11.2"
+```
+然后执行
+```bat
+msbuild infer_demo.sln /m:4 /p:Configuration=Release /p:Platform=x64
+```
+编译好的exe保存在Release目录下，在运行demo前，需要将模型和测试图片拷贝至该目录。另外，需要在终端指定DLL的搜索路径。请在build目录下执行以下命令。
+```bat
+set FASTDEPLOY_PATH=%cd%\..\..\..\..\..\..\..\fastdeploy-win-x64-gpu-0.2.0
+set PATH=%FASTDEPLOY_PATH%\lib;%FASTDEPLOY_PATH%\third_libs\install\onnxruntime\lib;%FASTDEPLOY_PATH%\third_libs\install\opencv-win-x64-3.4.16\build\x64\vc15\bin;%FASTDEPLOY_PATH%\third_libs\install\paddle_inference\paddle\lib;%FASTDEPLOY_PATH%\third_libs\install\paddle_inference\third_party\install\mkldnn\lib;%FASTDEPLOY_PATH%\third_libs\install\paddle_inference\third_party\install\mklml\lib;%FASTDEPLOY_PATH%\third_libs\install\paddle2onnx\lib;%FASTDEPLOY_PATH%\third_libs\install\tensorrt\lib;%FASTDEPLOY_PATH%\third_libs\install\yaml-cpp\lib;%PATH%
+```
+注意，需要拷贝onnxruntime.dll到exe所在的目录。
+```bat
+copy /Y %FASTDEPLOY_PATH%\third_libs\install\onnxruntime\lib\onnxruntime* Release\
+```
+由于较新的Windows在System32系统目录下自带了onnxruntime.dll，因此就算设置了PATH，系统依然会出现onnxruntime的加载冲突。因此需要先拷贝demo用到的onnxruntime.dll到exe所在的目录。
+```bat
+where onnxruntime.dll
+C:\Windows\System32\onnxruntime.dll  # windows自带的onnxruntime.dll
+```
+## 运行 demo
+```bat
+cd Release
+infer_ppyoloe_demo.exe ppyoloe_crn_l_300e_coco 000000014439.jpg 0  # CPU
+infer_ppyoloe_demo.exe ppyoloe_crn_l_300e_coco 000000014439.jpg 1  # GPU
+infer_ppyoloe_demo.exe ppyoloe_crn_l_300e_coco 000000014439.jpg 2  # GPU + TensorRT
+```
--- a/docs/develop.md
+++ b/docs/develop.md
@@ -1,7 +1,7 @@
 # 代码提交说明

 FastDeploy使用clang-format, cpplint检查和格式化代码，提交代码前，需安装pre-commit
-```
+```bash
 git clone https://github.com/PaddlePaddle/FastDeploy.git
 cd FastDeploy
 git checkout develop
--- a/docs/quick_start/Python_prebuilt_wheels.md
+++ b/docs/quick_start/Python_prebuilt_wheels.md
@@ -9,11 +9,11 @@ FastDeploy提供了在Windows/Linux/Mac上的预先编译Python Wheel包，开
 - Mac 支持Python3.6~3.9

 ## 安装 CPU Python 版本
-```
+```bash
 pip install fastdeploy-python -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html
 ```
 ## 安装 GPU Python 版本
-```
+```bash
 pip install fastdeploy-gpu-python -f https://www.paddlepaddle.org.cn/whl/fastdeploy.html
 ```

--- a/examples/vision/classification/paddleclas/cpp/README.md
+++ b/examples/vision/classification/paddleclas/cpp/README.md
@@ -9,14 +9,14 @@

 以Linux上ResNet50_vd推理为例，在本目录执行如下命令即可完成编译测试

-```
+```bash
 #下载SDK，编译模型examples代码（SDK中包含了examples代码）
 wget https://bj.bcebos.com/paddlehub/fastdeploy/libs/0.2.0/fastdeploy-linux-x64-gpu-0.2.0.tgz
 tar xvf fastdeploy-linux-x64-gpu-0.2.0.tgz
 cd fastdeploy-linux-x64-gpu-0.2.0/examples/vision/classification/paddleclas/cpp
 mkdir build
 cd build
-cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/../../../../../../../fastdeploy-linux-x64-gpu-0.2.0 
+cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/../../../../../../../fastdeploy-linux-x64-gpu-0.2.0
 make -j

 # 下载ResNet50_vd模型文件和测试图片
@@ -33,11 +33,14 @@ wget https://gitee.com/paddlepaddle/PaddleClas/raw/release/2.4/deploy/images/Ima
 ./infer_demo ResNet50_vd_infer ILSVRC2012_val_00000010.jpeg 2
 ```

+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)
+
 ## PaddleClas C++接口

 ### PaddleClas类

-```
+```c++
 fastdeploy::vision::classification::PaddleClasModel(
        const string& model_file,
        const string& params_file,
@@ -58,14 +61,14 @@ PaddleClas模型加载和初始化，其中model_file, params_file为训练模

 #### Predict函数

-> ```
+> ```c++
 > PaddleClasModel::Predict(cv::Mat* im, ClassifyResult* result, int topk = 1)
 > ```
-> 
+>
 > 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**: 分类结果，包括label_id，以及相应的置信度, ClassifyResult说明参考[视觉模型预测结果](../../../../../docs/api/vision_results/)
 > > * **topk**(int):返回预测概率最高的topk个分类结果，默认为1
--- a/examples/vision/classification/paddleclas/python/README.md
+++ b/examples/vision/classification/paddleclas/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成ResNet50_vd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  FastDeploy/examples/vision/classification/paddleclas/python
@@ -26,7 +26,7 @@ python infer.py --model ResNet50_vd_infer --image ILSVRC2012_val_00000010.jpeg -
 ```

 运行完成后返回结果如下所示
-```
+```bash
 ClassifyResult(
 label_ids: 153,
 scores: 0.686229,
@@ -35,7 +35,7 @@ scores: 0.686229,

 ## PaddleClasModel Python接口

-```
+```python
 fd.vision.classification.PaddleClasModel(model_file, params_file, config_file, runtime_option=None, model_format=Frontend.PADDLE)
 ```

@@ -51,19 +51,19 @@ PaddleClas模型加载和初始化，其中model_file, params_file为训练模

 ### predict函数

-> ```
+> ```python
 > PaddleClasModel.predict(input_image, topk=1)
 > ```
-> 
+>
 > 模型预测结口，输入图像直接输出检测结果。
-> 
+>
 > **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/)


--- a/examples/vision/detection/nanodet_plus/cpp/README.md
+++ b/examples/vision/detection/nanodet_plus/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301689-87ee5205-2eff-4204-b615-24c400f01323.jpg">

+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)
+
 ## NanoDetPlus C++接口

 ### NanoDetPlus类

-```
+```c++
 fastdeploy::vision::detection::NanoDetPlus(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ NanoDetPlus模型加载和初始化，其中model_file为导出的ONNX模型格

 #### Predict函数

-> ```
+> ```c++
 > NanoDetPlus::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/nanodet_plus/python/README.md
+++ b/examples/vision/detection/nanodet_plus/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成NanoDetPlus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/detection/nanodet_plus/python/
@@ -30,7 +30,7 @@ python infer.py --model nanodet-plus-m_320.onnx --image 000000014439.jpg --devic

 ## NanoDetPlus Python接口

-```
+```python
 fastdeploy.vision.detection.NanoDetPlus(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ NanoDetPlus模型加载和初始化，其中model_file为导出的ONNX模型格

 ### predict函数

-> ```
+> ```python
 > NanoDetPlus.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/paddledetection/cpp/README.md
+++ b/examples/vision/detection/paddledetection/cpp/README.md
@@ -9,7 +9,7 @@

 以Linux上推理为例，在本目录执行如下命令即可完成编译测试

-```
+```bash
 以ppyoloe为例进行推理部署

 #下载SDK，编译模型examples代码（SDK中包含了examples代码）
@@ -34,12 +34,15 @@ tar xvf ppyoloe_crn_l_300e_coco.tgz
 ./infer_ppyoloe_demo ./ppyoloe_crn_l_300e_coco 000000014439.jpg 2
 ```

+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)
+
 ## PaddleDetection C++接口

 ### 模型类

 PaddleDetection目前支持6种模型系列，类名分别为`PPYOLOE`, `PicoDet`, `PaddleYOLOX`, `PPYOLO`, `FasterRCNN`，所有类名的构造函数和预测函数在参数上完全一致，本文档以PPYOLOE为例讲解API
-```
+```c++
 fastdeploy::vision::detection::PPYOLOE(
        const string& model_file,
        const string& params_file,
@@ -60,7 +63,7 @@ PaddleDetection PPYOLOE模型加载和初始化，其中model_file为导出的ON

 #### Predict函数

-> ```
+> ```c++
 > PPYOLOE::Predict(cv::Mat* im, DetectionResult* result)
 > ```
 >
--- a/examples/vision/detection/paddledetection/python/README.md
+++ b/examples/vision/detection/paddledetection/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer_xxx.py`快速完成PPYOLOE/PicoDet等模型在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/detection/paddledetection/python/
@@ -26,13 +26,13 @@ python infer_ppyoloe.py --model_dir ppyoloe_crn_l_300e_coco --image 000000014439
 ```

 运行完成可视化结果如下图所示
-<div  align="center">    
+<div  align="center">  
 <img src="https://user-images.githubusercontent.com/19339784/184326520-7075e907-10ed-4fad-93f8-52d0e35d4964.jpg", width=480px, height=320px />
 </div>

 ## PaddleDetection Python接口

-```
+```python
 fastdeploy.vision.detection.PPYOLOE(model_file, params_file, config_file, runtime_option=None, model_format=Frontend.PADDLE)
 fastdeploy.vision.detection.PicoDet(model_file, params_file, config_file, runtime_option=None, model_format=Frontend.PADDLE)
 fastdeploy.vision.detection.PaddleYOLOX(model_file, params_file, config_file, runtime_option=None, model_format=Frontend.PADDLE)
@@ -54,7 +54,7 @@ PaddleDetection模型加载和初始化，其中model_file， params_file为导
 ### predict函数

 PaddleDetection中各个模型，包括PPYOLOE/PicoDet/PaddleYOLOX/YOLOv3/PPYOLO/FasterRCNN，均提供如下同样的成员函数用于进行图像的检测
-> ```
+> ```python
 > PPYOLOE.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/scaledyolov4/README.md
+++ b/examples/vision/detection/scaledyolov4/README.md
@@ -11,7 +11,7 @@

  访问[ScaledYOLOv4](https://github.com/WongKinYiu/ScaledYOLOv4)官方github库，按照指引下载安装，下载`scaledyolov4.pt` 模型，利用 `models/export.py` 得到`onnx`格式文件。如果您导出的`onnx`模型出现问题，可以参考[ScaledYOLOv4#401](https://github.com/WongKinYiu/ScaledYOLOv4/issues/401)的解决办法

-  ```
+  ```bash
  #下载ScaledYOLOv4模型文件
  Download from the goole drive https://drive.google.com/file/d/1aXZZE999sHMP1gev60XhNChtHPRMH3Fz/view?usp=sharing

--- a/examples/vision/detection/scaledyolov4/cpp/README.md
+++ b/examples/vision/detection/scaledyolov4/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301908-7027cf41-af51-4485-bd32-87aca0e77336.jpg">

+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)
+
 ## ScaledYOLOv4 C++接口

 ### ScaledYOLOv4类

-```
+```c++
 fastdeploy::vision::detection::ScaledYOLOv4(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ ScaledYOLOv4模型加载和初始化，其中model_file为导出的ONNX模型格

 #### Predict函数

-> ```
+> ```c++
 > ScaledYOLOv4::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/scaledyolov4/python/README.md
+++ b/examples/vision/detection/scaledyolov4/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成ScaledYOLOv4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/detection/scaledyolov4/python/
@@ -30,7 +30,7 @@ python infer.py --model scaled_yolov4-p5.onnx --image 000000014439.jpg --device

 ## ScaledYOLOv4 Python接口

-```
+```python
 fastdeploy.vision.detection.ScaledYOLOv4(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ ScaledYOLOv4模型加载和初始化，其中model_file为导出的ONNX模型格

 ### predict函数

-> ```
+> ```python
 > ScaledYOLOv4.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/yolor/README.md
+++ b/examples/vision/detection/yolor/README.md
@@ -11,7 +11,7 @@

  访问[YOLOR](https://github.com/WongKinYiu/yolor)官方github库，按照指引下载安装，下载`yolor.pt` 模型，利用 `models/export.py` 得到`onnx`格式文件。如果您导出的`onnx`模型出现精度不达标或者是数据维度的问题，可以参考[yolor#32](https://github.com/WongKinYiu/yolor/issues/32)的解决办法

-  ```
+  ```bash
  #下载yolor模型文件
  wget https://github.com/WongKinYiu/yolor/releases/download/weights/yolor-d6-paper-570.pt

--- a/examples/vision/detection/yolor/cpp/README.md
+++ b/examples/vision/detection/yolor/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301926-fa3711bf-5984-4e61-9c98-7fdeacb622e9.jpg">

+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)
+
 ## YOLOR C++接口

 ### YOLOR类

-```
+```c++
 fastdeploy::vision::detection::YOLOR(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOR模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > YOLOR::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/yolor/python/README.md
+++ b/examples/vision/detection/yolor/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOR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/detection/yolor/python/
@@ -30,7 +30,7 @@ python infer.py --model yolor-p6-paper-541-640-640.onnx --image 000000014439.jpg

 ## YOLOR Python接口

-```
+```python
 fastdeploy.vision.detection.YOLOR(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ YOLOR模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > YOLOR.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/yolov5/cpp/README.md
+++ b/examples/vision/detection/yolov5/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184309358-d803347a-8981-44b6-b589-4608021ad0f4.jpg">

+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)
+
 ## YOLOv5 C++接口

 ### YOLOv5类

-```
+```c++
 fastdeploy::vision::detection::YOLOv5(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOv5模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > YOLOv5::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/yolov5/python/README.md
+++ b/examples/vision/detection/yolov5/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOv5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/detection/yolov5/python/
@@ -30,7 +30,7 @@ python infer.py --model yolov5s.onnx --image 000000014439.jpg --device gpu --use

 ## YOLOv5 Python接口

-```
+```python
 fastdeploy.vision.detection.YOLOv5(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ YOLOv5模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > YOLOv5.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/yolov5lite/README.md
+++ b/examples/vision/detection/yolov5lite/README.md
@@ -12,7 +12,7 @@
 - 自动获取
  访问[YOLOv5Lite](https://github.com/ppogg/YOLOv5-Lite)
 官方github库，按照指引下载安装，下载`yolov5-lite-xx.onnx` 模型(Tips：官方提供的ONNX文件目前是没有decode模块的)
-  ```
+  ```bash
  #下载yolov5-lite模型文件(.onnx)
  Download from https://drive.google.com/file/d/1bJByk9eoS6pv8Z3N4bcLRCV3i7uk24aU/view
  官方Repo也支持百度云下载
@@ -27,7 +27,7 @@

  首先需要参考[YOLOv5-Lite#189](https://github.com/ppogg/YOLOv5-Lite/pull/189)的解决办法，修改代码。

-  ```
+  ```bash
  #下载yolov5-lite模型文件(.pt)
  Download from https://drive.google.com/file/d/1oftzqOREGqDCerf7DtD5BZp9YWELlkMe/view
  官方Repo也支持百度云下载
@@ -39,7 +39,7 @@
  ```
  - 导出无decode模块的ONNX文件(不需要修改代码)

-  ```
+  ```bash
  #下载yolov5-lite模型文件
  Download from https://drive.google.com/file/d/1oftzqOREGqDCerf7DtD5BZp9YWELlkMe/view
  官方Repo也支持百度云下载
--- a/examples/vision/detection/yolov5lite/cpp/README.md
+++ b/examples/vision/detection/yolov5lite/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301943-263c8153-a52a-4533-a7c1-ee86d05d314b.jpg">

+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)
+
 ## YOLOv5Lite C++接口

 ### YOLOv5Lite类

-```
+```c++
 fastdeploy::vision::detection::YOLOv5Lite(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOv5Lite模型加载和初始化，其中model_file为导出的ONNX模型格

 #### Predict函数

-> ```
+> ```c++
 > YOLOv5Lite::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/yolov5lite/python/README.md
+++ b/examples/vision/detection/yolov5lite/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOv5Lite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/detection/yolov5lite/python/
@@ -30,7 +30,7 @@ python infer.py --model v5Lite-g-sim-640.onnx --image 000000014439.jpg --device

 ## YOLOv5Lite Python接口

-```
+```python
 fastdeploy.vision.detection.YOLOv5Lite(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ YOLOv5Lite模型加载和初始化，其中model_file为导出的ONNX模型格

 ### predict函数

-> ```
+> ```python
 > YOLOv5Lite.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/yolov6/cpp/README.md
+++ b/examples/vision/detection/yolov6/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301725-390e4abb-db2b-482d-931d-469381322626.jpg">

+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)
+
 ## YOLOv6 C++接口

 ### YOLOv6类

-```
+```c++
 fastdeploy::vision::detection::YOLOv6(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOv6模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > YOLOv6::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/yolov6/python/README.md
+++ b/examples/vision/detection/yolov6/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOv6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/detection/yolov6/python/
@@ -31,7 +31,7 @@ python infer.py --model yolov6s.onnx --image 000000014439.jpg --device gpu --use

 ## YOLOv6 Python接口

-```
+```python
 fastdeploy.vision.detection.YOLOv6(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -46,7 +46,7 @@ YOLOv6模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > YOLOv6.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/yolov7/README.md
+++ b/examples/vision/detection/yolov7/README.md
@@ -10,7 +10,7 @@

 ## 导出ONNX模型

-```
+```bash
 # 下载yolov7模型文件
 wget https://github.com/WongKinYiu/yolov7/releases/download/v0.1/yolov7.pt

--- a/examples/vision/detection/yolov7/cpp/README.md
+++ b/examples/vision/detection/yolov7/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/183847558-abcd9a57-9cd9-4891-b09a-710963c99b74.jpg">

+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)
+
 ## YOLOv7 C++接口

 ### YOLOv7类

-```
+```c++
 fastdeploy::vision::detection::YOLOv7(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOv7模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > YOLOv7::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/yolov7/python/README.md
+++ b/examples/vision/detection/yolov7/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOv7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/detection/yolov7/python/
@@ -30,7 +30,7 @@ python infer.py --model yolov7.onnx --image 000000014439.jpg --device gpu --use_

 ## YOLOv7 Python接口

-```
+```python
 fastdeploy.vision.detection.YOLOv7(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ YOLOv7模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > YOLOv7.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/detection/yolox/cpp/README.md
+++ b/examples/vision/detection/yolox/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://gitee.com/paddlepaddle/PaddleDetection/raw/release/2.4/demo/0000000

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301746-04595d76-454a-4f07-8c7d-6f41418f8ae3.jpg">

+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)
+
 ## YOLOX C++接口

 ### YOLOX类

-```
+```c++
 fastdeploy::vision::detection::YOLOX(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOX模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > YOLOX::Predict(cv::Mat* im, DetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/detection/yolox/python/README.md
+++ b/examples/vision/detection/yolox/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOX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/detection/yolox/python/
@@ -30,7 +30,7 @@ python infer.py --model yolox_s.onnx --image 000000014439.jpg --device gpu --use

 ## YOLOX Python接口

-```
+```python
 fastdeploy.vision.detection.YOLOX(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ YOLOX模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > YOLOX.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/facedet/retinaface/cpp/README.md
+++ b/examples/vision/facedet/retinaface/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,13 @@ wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/li

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301763-1b950047-c17f-4819-b175-c743b699c3b1.jpg">

+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)
 ## RetinaFace C++接口

 ### RetinaFace类

-```
+```c++
 fastdeploy::vision::facedet::RetinaFace(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +59,7 @@ RetinaFace模型加载和初始化，其中model_file为导出的ONNX模型格

 #### Predict函数

-> ```
+> ```c++
 > RetinaFace::Predict(cv::Mat* im, FaceDetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/facedet/retinaface/python/README.md
+++ b/examples/vision/facedet/retinaface/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成RetinaFace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//retinaface/python/
@@ -30,7 +30,7 @@ python infer.py --model Pytorch_RetinaFace_mobile0.25-640-640.onnx --image test_

 ## RetinaFace Python接口

-```
+```python
 fastdeploy.vision.facedet.RetinaFace(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ RetinaFace模型加载和初始化，其中model_file为导出的ONNX模型格

 ### predict函数

-> ```
+> ```python
 > RetinaFace.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/facedet/scrfd/README.md
+++ b/examples/vision/facedet/scrfd/README.md
@@ -8,7 +8,7 @@

 ## 导出ONNX模型

-  ```
+  ```bash
  #下载scrfd模型文件
  e.g. download from  https://onedrive.live.com/?authkey=%21ABbFJx2JMhNjhNA&id=4A83B6B633B029CC%215542&cid=4A83B6B633B029CC

--- a/examples/vision/facedet/scrfd/cpp/README.md
+++ b/examples/vision/facedet/scrfd/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/li

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301789-1981d065-208f-4a6b-857c-9a0f9a63e0b1.jpg">

+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)
+
 ## SCRFD C++接口

 ### SCRFD类

-```
+```c++
 fastdeploy::vision::facedet::SCRFD(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ SCRFD模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > SCRFD::Predict(cv::Mat* im, FaceDetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/facedet/scrfd/python/README.md
+++ b/examples/vision/facedet/scrfd/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成SCRFD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/facedet/scrfd/python/
@@ -30,7 +30,7 @@ python infer.py --model scrfd_500m_bnkps_shape640x640.onnx --image test_lite_fac

 ## SCRFD Python接口

-```
+```python
 fastdeploy.vision.facedet.SCRFD(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ SCRFD模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > SCRFD.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/facedet/ultraface/cpp/README.md
+++ b/examples/vision/facedet/ultraface/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/li

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301821-0788483b-a72b-42b0-a566-b6430f184f6e.jpg">

+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)
+
 ## UltraFace C++接口

 ### UltraFace类

-```
+```c++
 fastdeploy::vision::facedet::UltraFace(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ UltraFace模型加载和初始化，其中model_file为导出的ONNX模型格式

 #### Predict函数

-> ```
+> ```c++
 > UltraFace::Predict(cv::Mat* im, FaceDetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/facedet/ultraface/python/README.md
+++ b/examples/vision/facedet/ultraface/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成UltraFace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/facedet/ultraface/python/
@@ -30,7 +30,7 @@ python infer.py --model version-RFB-320.onnx --image test_lite_face_detector_3.j

 ## UltraFace Python接口

-```
+```python
 fastdeploy.vision.facedet.UltraFace(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ UltraFace模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > UltraFace.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/facedet/yolov5face/cpp/README.md
+++ b/examples/vision/facedet/yolov5face/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -34,11 +34,14 @@ wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/li

 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301839-a29aefae-16c9-4196-bf9d-9c6cf694f02d.jpg">

+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)
+
 ## YOLOv5Face C++接口

 ### YOLOv5Face类

-```
+```c++
 fastdeploy::vision::facedet::YOLOv5Face(
        const string& model_file,
        const string& params_file = "",
@@ -57,7 +60,7 @@ YOLOv5Face模型加载和初始化，其中model_file为导出的ONNX模型格

 #### Predict函数

-> ```
+> ```c++
 > YOLOv5Face::Predict(cv::Mat* im, FaceDetectionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/facedet/yolov5face/python/README.md
+++ b/examples/vision/facedet/yolov5face/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成YOLOv5Face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/facedet/yolov5face/python/
@@ -30,7 +30,7 @@ python infer.py --model yolov5s-face.onnx --image test_lite_face_detector_3.jpg

 ## YOLOv5Face Python接口

-```
+```python
 fastdeploy.vision.facedet.YOLOv5Face(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -45,7 +45,7 @@ YOLOv5Face模型加载和初始化，其中model_file为导出的ONNX模型格

 ### predict函数

-> ```
+> ```python
 > YOLOv5Face.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/faceid/insightface/README.md
+++ b/examples/vision/faceid/insightface/README.md
@@ -18,7 +18,7 @@
 访问[ArcFace](https://github.com/deepinsight/insightface/tree/master/recognition/arcface_torch)官方github库，按照指引下载安装，下载pt模型文件，利用 `torch2onnx.py` 得到`onnx`格式文件。

 * 下载ArcFace模型文件
-  ```
+  ```bash
  Link: https://pan.baidu.com/share/init?surl=CL-l4zWqsI1oDuEEYVhj-g code: e8pw  
  ```

--- a/examples/vision/faceid/insightface/cpp/README.md
+++ b/examples/vision/faceid/insightface/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -40,11 +40,14 @@ wget https://bj.bcebos.com/paddlehub/test_samples/test_lite_focal_arcface_2.JPG
 <img width="220" float="left" src="https://user-images.githubusercontent.com/67993288/184321622-d9a494c3-72f3-47f1-97c5-8a2372de491f.JPG">
 </div>

+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)
+
 ## InsightFace C++接口

 ### ArcFace类

-```
+```c++
 fastdeploy::vision::faceid::ArcFace(
        const string& model_file,
        const string& params_file = "",
@@ -56,7 +59,7 @@ ArcFace模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### CosFace类

-```
+```c++
 fastdeploy::vision::faceid::CosFace(
        const string& model_file,
        const string& params_file = "",
@@ -68,7 +71,7 @@ CosFace模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### PartialFC类

-```
+```c++
 fastdeploy::vision::faceid::PartialFC(
        const string& model_file,
        const string& params_file = "",
@@ -80,7 +83,7 @@ PartialFC模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### VPL类

-```
+```c++
 fastdeploy::vision::faceid::VPL(
        const string& model_file,
        const string& params_file = "",
@@ -98,7 +101,7 @@ VPL模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > ArcFace::Predict(cv::Mat* im, FaceRecognitionResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/faceid/insightface/python/README.md
+++ b/examples/vision/faceid/insightface/python/README.md
@@ -8,7 +8,7 @@

 以ArcFace为例子, 提供`infer_arcface.py`快速完成ArcFace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/faceid/insightface/python/
@@ -35,7 +35,7 @@ python infer_arcface.py --model ms1mv3_arcface_r100.onnx --face test_lite_focal_
 <img width="220" float="left" src="https://user-images.githubusercontent.com/67993288/184321622-d9a494c3-72f3-47f1-97c5-8a2372de491f.JPG">
 </div>

-```
+```bash
 Prediction Done!
 --- [Face 0]:FaceRecognitionResult: [Dim(512), Min(-2.309220), Max(2.372197), Mean(0.016987)]
 --- [Face 1]:FaceRecognitionResult: [Dim(512), Min(-2.288258), Max(1.995104), Mean(-0.003400)]
@@ -46,7 +46,7 @@ Detect Done! Cosine 01: 0.814385, Cosine 02:-0.059388

 ## InsightFace Python接口

-```
+```python
 fastdeploy.vision.faceid.ArcFace(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 fastdeploy.vision.faceid.CosFace(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 fastdeploy.vision.faceid.PartialFC(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
@@ -64,7 +64,7 @@ ArcFace模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > ArcFace.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/matting/modnet/cpp/README.md
+++ b/examples/vision/matting/modnet/cpp/README.md
@@ -9,7 +9,7 @@

 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://https://bj.bcebos.com/paddlehub/fastdeploy/cpp/fastdeploy-linux-x64-gpu-0.2.0.tgz
@@ -36,11 +36,14 @@ wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/li
 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301892-457f7014-2dc0-4ad1-b688-43b41fac299a.jpg">
 <img width="640" src="https://user-images.githubusercontent.com/67993288/184301871-c234dfdf-3b3d-46e4-8886-e1ac156c9e4a.jpg">

+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)
+
 ## MODNet C++接口

 ### MODNet类

-```
+```c++
 fastdeploy::vision::matting::MODNet(
        const string& model_file,
        const string& params_file = "",
@@ -59,7 +62,7 @@ MODNet模型加载和初始化，其中model_file为导出的ONNX模型格式。

 #### Predict函数

-> ```
+> ```c++
 > MODNet::Predict(cv::Mat* im, MattingResult* result,
 >                 float conf_threshold = 0.25,
 >                 float nms_iou_threshold = 0.5)
--- a/examples/vision/matting/modnet/python/README.md
+++ b/examples/vision/matting/modnet/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成MODNet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/matting/modnet/python/
@@ -31,7 +31,7 @@ python infer.py --model modnet_photographic_portrait_matting.onnx --image test_l

 ## MODNet Python接口

-```
+```python
 fastdeploy.vision.matting.MODNet(model_file, params_file=None, runtime_option=None, model_format=Frontend.ONNX)
 ```

@@ -46,7 +46,7 @@ MODNet模型加载和初始化，其中model_file为导出的ONNX模型格式

 ### predict函数

-> ```
+> ```python
 > MODNet.predict(image_data, conf_threshold=0.25, nms_iou_threshold=0.5)
 > ```
 >
--- a/examples/vision/segmentation/paddleseg/cpp/README.md
+++ b/examples/vision/segmentation/paddleseg/cpp/README.md
@@ -9,13 +9,13 @@

 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/paddlehub/fastdeploy/libs/0.2.0/fastdeploy-linux-x64-gpu-0.2.0.tgz
 tar xvf fastdeploy-linux-x64-gpu-0.2.0.tgz
 cd fastdeploy-linux-x64-gpu-0.2.0/examples/vision/segmentation/paddleseg/cpp/build
-cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/../../../../../../../fastdeploy-linux-x64-gpu-0.2.0 
+cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/../../../../../../../fastdeploy-linux-x64-gpu-0.2.0
 make -j

 # 下载Unet模型文件和测试图片
@@ -33,15 +33,18 @@ wget https://paddleseg.bj.bcebos.com/dygraph/demo/cityscapes_demo.png
 ```

 运行完成可视化结果如下图所示
-<div  align="center">    
+<div  align="center">  
 <img src="https://user-images.githubusercontent.com/16222477/184588768-45ee673b-ef1f-40f4-9fbd-6b1a9ce17c59.png", width=512px, height=256px />
 </div>

+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)
+
 ## PaddleSeg C++接口

 ### PaddleSeg类

-```
+```c++
 fastdeploy::vision::segmentation::PaddleSegModel(
        const string& model_file,
        const string& params_file = "",
@@ -62,7 +65,7 @@ PaddleSegModel模型加载和初始化，其中model_file为导出的Paddle模

 #### Predict函数

-> ```
+> ```c++
 > PaddleSegModel::Predict(cv::Mat* im, DetectionResult* result)
 > ```
 >
--- a/examples/vision/segmentation/paddleseg/python/README.md
+++ b/examples/vision/segmentation/paddleseg/python/README.md
@@ -7,7 +7,7 @@

 本目录下提供`infer.py`快速完成Unet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 FastDeploy/examples/vision/segmentation/paddleseg/python
@@ -26,13 +26,13 @@ python infer.py --model Unet_cityscapes_without_argmax_infer --image cityscapes_
 ```

 运行完成可视化结果如下图所示
-<div  align="center">    
+<div  align="center">  
 <img src="https://user-images.githubusercontent.com/16222477/184588768-45ee673b-ef1f-40f4-9fbd-6b1a9ce17c59.png", width=512px, height=256px />
 </div>

 ## PaddleSegModel Python接口

-```
+```python
 fd.vision.segmentation.PaddleSegModel(model_file, params_file, config_file, runtime_option=None, model_format=Frontend.PADDLE)
 ```

@@ -48,7 +48,7 @@ PaddleSeg模型加载和初始化，其中model_file, params_file以及config_fi

 ### predict函数

-> ```
+> ```python
 > PaddleSegModel.predict(input_image)
 > ```
 >
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,3 +1,4 @@
+wheel
 requests
 tqdm
 numpy