[Model] Support BlazeFace Model (PaddlePaddle#1172)

* fit yolov7face file path * TODO：添加yolov7facePython接口Predict * resolve yolov7face.py * resolve yolov7face.py * resolve yolov7face.py * add yolov7face example readme file * [Doc] fix yolov7face example readme file * [Doc]fix yolov7face example readme file * support BlazeFace * add blazeface readme file * fix review problem * fix code style error * fix review problem * fix review problem * fix head file problem * fix review problem * fix review problem * fix readme file problem * add English readme file * fix English readme file
heliqi · Feb 6, 2023 · 42d14e7 · 42d14e7
1 parent e2de3f3
commit 42d14e7
Show file tree

Hide file tree

Showing 21 changed files with 1,518 additions and 0 deletions.
diff --git a/examples/vision/facedet/blazeface/README.md b/examples/vision/facedet/blazeface/README.md
@@ -0,0 +1,34 @@
+English | [简体中文](README_CN.md)
+# BlazeFace Ready-to-deploy Model
+
+- BlazeFace deployment model implementation comes from [BlazeFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection),and [Pre-training model based on WiderFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection)
+  - （1）Provided in [Official library
+](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/tools) *.params, could deploy after operation [export_model.py](#Export PADDLE model);
+  - （2）Developers can train BlazeFace model based on their own data according to [export_model. py](https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.5/tools/export_model.py)After exporting the model, complete the deployment。
+
+## Export PADDLE model
+
+Visit [BlazeFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection) Github library, download and install according to the instructions, download the `. yml` and `. params` model parameters, and use` export_ Model. py `gets the` pad `model file`. yml,. pdiparams,. pdmodel `.
+
+
+* Download BlazeFace model parameter file
+
+|Network structure | input size | number of pictures/GPU | learning rate strategy | Easy/Media/Hard Set | prediction delay (SD855) | model size (MB) | download | configuration file|
+|:------------:|:--------:|:----:|:-------:|:-------:|:---------:|:----------:|:---------:|:--------:|
+| BlazeFace  | 640  |    8    | 1000e     | 0.885 / 0.855 / 0.731 | - | 0.472 |[Download link](https://paddledet.bj.bcebos.com/models/blazeface_1000e.pdparams) | [Config file](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection/blazeface_1000e.yml) |
+| BlazeFace-FPN-SSH  | 640  |    8    | 1000e     | 0.907 / 0.883 / 0.793 | - | 0.479 |[Download link](https://paddledet.bj.bcebos.com/models/blazeface_fpn_ssh_1000e.pdparams) | [Config file](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection/blazeface_fpn_ssh_1000e.yml) |
+
+* Export paddle-format file
+  ```bash
+  python tools/export_model.py -c configs/face_detection/blazeface_1000e.yml -o weights=blazeface_1000e.pdparams --export_serving_model=True
+  ```
+
+## Detailed Deployment Tutorials
+
+- [Python Deployment](python)
+- [C++ Deployment](cpp)
+
+
+## Release Note
+
+- This tutorial and related code are written based on [BlazeFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection)
diff --git a/examples/vision/facedet/blazeface/README_CN.md b/examples/vision/facedet/blazeface/README_CN.md
@@ -0,0 +1,31 @@
+# BlazeFace准备部署模型
+
+- BlazeFace部署模型实现来自[BlazeFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection),和[基于WiderFace的预训练模型](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection)
+  - （1）[官方库](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/tools)中提供的*.params,通过[export_model.py](#导出PADDLE模型)操作后，可进行部署；
+  - （2）开发者基于自己数据训练的BlazeFace模型，可按照[export_model.py](https://github.com/PaddlePaddle/PaddleDetection/blob/release/2.5/tools/export_model.py)导出模型后，完成部署。
+
+## 导出PADDLE模型
+
+访问[BlazeFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection)github库，按照指引下载安装，下载`.yml`和`.params` 模型参数，利用 `export_model.py` 得到`paddle`模型文件`.yml, .pdiparams, .pdmodel`。
+
+* 下载BlazeFace模型参数文件
+
+| 网络结构 | 输入尺寸 | 图片个数/GPU | 学习率策略 | Easy/Medium/Hard Set  | 预测时延（SD855）| 模型大小(MB) | 下载 | 配置文件 |
+|:------------:|:--------:|:----:|:-------:|:-------:|:---------:|:----------:|:---------:|:--------:|
+| BlazeFace  | 640  |    8    | 1000e     | 0.885 / 0.855 / 0.731 | - | 0.472 |[下载链接](https://paddledet.bj.bcebos.com/models/blazeface_1000e.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection/blazeface_1000e.yml) |
+| BlazeFace-FPN-SSH  | 640  |    8    | 1000e     | 0.907 / 0.883 / 0.793 | - | 0.479 |[下载链接](https://paddledet.bj.bcebos.com/models/blazeface_fpn_ssh_1000e.pdparams) | [配置文件](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection/blazeface_fpn_ssh_1000e.yml) |
+
+* 导出paddle格式文件
+  ```bash
+  python tools/export_model.py -c configs/face_detection/blazeface_1000e.yml -o weights=blazeface_1000e.pdparams --export_serving_model=True
+  ```
+
+## 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[BlazeFace](https://github.com/PaddlePaddle/PaddleDetection/tree/release/2.5/configs/face_detection) 编写
diff --git a/examples/vision/facedet/blazeface/cpp/CMakeLists.txt b/examples/vision/facedet/blazeface/cpp/CMakeLists.txt
@@ -0,0 +1,14 @@
+PROJECT(infer_demo C CXX)
+CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
+
+# Specifies the path to the fastdeploy library after you have downloaded it
+option(FASTDEPLOY_INSTALL_DIR "Path of downloaded fastdeploy sdk.")
+
+include(../../../../../FastDeploy.cmake)
+
+# Add the FastDeploy dependency header
+include_directories(${FASTDEPLOY_INCS})
+
+add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer.cc)
+# Add the FastDeploy library dependency
+target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
diff --git a/examples/vision/facedet/blazeface/cpp/README.md b/examples/vision/facedet/blazeface/cpp/README.md
@@ -0,0 +1,78 @@
+English | [简体中文](README_CN.md)
+# BlazeFace C++ Deployment Example
+
+This directory provides examples that `infer.cc` fast finishes the deployment of BlazeFace on CPU/GPU。
+
+Before deployment, two steps require confirmation
+
+- 1. Software and hardware should meet the requirements. Please refer to [FastDeploy Environment Requirements](../../../../../docs/en/build_and_install/download_prebuilt_libraries.md)  
+- 2. Download the precompiled deployment library and samples code according to your development environment. Refer to [FastDeploy Precompiled Library](../../../../../docs/en/build_and_install/download_prebuilt_libraries.md)
+
+Taking the CPU inference on Linux as an example, the compilation test can be completed by executing the following command in this directory.
+
+```bash
+mkdir build
+cd build
+# Download the FastDeploy precompiled library. Users can choose your appropriate version in the `FastDeploy Precompiled Library` mentioned above 
+wget https://bj.bcebos.com/fastdeploy/release/cpp/fastdeploy-linux-x64-x.x.x.tgz # x.x.x >= 1.0.4
+tar xvf fastdeploy-linux-x64-x.x.x.tgz # x.x.x >= 1.0.4
+cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/fastdeploy-linux-x64-x.x.x # x.x.x >= 1.0.4
+make -j
+
+#Download the official converted YOLOv7Face model files and test images
+wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/lite/resources/test_lite_face_detector_3.jpg
+wget https://bj.bcebos.com/paddlehub/fastdeploy/blzeface-1000e.tgz
+
+#Use blazeface-1000e model
+# CPU inference
+./infer_demo blazeface-1000e/ test_lite_face_detector_3.jpg 0
+# GPU Inference
+./infer_demo blazeface-1000e/ test_lite_face_detector_3.jpg 1
+```
+
+The visualized result after running is as follows
+
+<img width="640" src="https://user-images.githubusercontent.com/49013063/206170111-843febb6-67d6-4c46-a121-d87d003bba21.jpg">
+
+The above command works for Linux or MacOS. For SDK use-pattern in Windows, refer to:
+- [How to use FastDeploy C++ SDK in Windows](../../../../../docs/cn/faq/use_sdk_on_windows.md)
+
+## BlazeFace C++ Interface
+
+### BlazeFace Class
+
+```c++
+fastdeploy::vision::facedet::BlazeFace(
+        const string& model_file,
+        const string& params_file = "",
+        const string& config_file = "",
+        const RuntimeOption& runtime_option = RuntimeOption(),
+        const ModelFormat& model_format = ModelFormat::PADDLE)
+```
+
+BlazeFace model loading and initialization, among which model_file is the exported PADDLE model format
+
+**Parameter**
+
+> * **model_file**(str): Model file path 
+> * **params_file**(str): Parameter file path. Only passing an empty string when the model is in PADDLE format
+> * **config_file**(str): Config file path. Only passing an empty string when the model is in PADDLE format
+> * **runtime_option**(RuntimeOption): Backend inference configuration. None by default, which is the default configuration
+> * **model_format**(ModelFormat): Model format. PADDLE format by default
+
+#### Predict Function
+
+> ```c++
+> BlazeFace::Predict(cv::Mat& im, FaceDetectionResult* result)
+> ```
+>
+> Model prediction interface. Input images and output detection results.
+>
+> **Parameter**
+>
+> > * **im**: Input images in HWC or BGR format
+> > * **result**: Detection results, including detection box and confidence of each box. Refer to [Vision Model Prediction Result](../../../../../docs/api/vision_results/) for FaceDetectionResult
+
+- [Model Description](../../)
+- [Python Deployment](../python)
+- [Vision Model Prediction Results](../../../../../docs/api/vision_results/)
diff --git a/examples/vision/facedet/blazeface/cpp/README_CN.md b/examples/vision/facedet/blazeface/cpp/README_CN.md
@@ -0,0 +1,77 @@
+[English](README.md) | 简体中文
+# BlazeFace C++部署示例
+
+本目录下提供`infer.cc`快速完成BlazeFace在CPU/GPU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/cn/build_and_install/download_prebuilt_libraries.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/cn/build_and_install/download_prebuilt_libraries.md)
+
+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
+# 下载FastDeploy预编译库，用户可在上文提到的`FastDeploy预编译库`中自行选择合适的版本使用
+wget https://bj.bcebos.com/fastdeploy/release/cpp/fastdeploy-linux-x64-x.x.x.tgz # x.x.x >= 1.0.4
+tar xvf fastdeploy-linux-x64-x.x.x.tgz # x.x.x >= 1.0.4
+cmake .. -DFASTDEPLOY_INSTALL_DIR=${PWD}/fastdeploy-linux-x64-x.x.x # x.x.x >= 1.0.4
+make -j
+
+#下载官方转换好的BlazeFace模型文件和测试图片
+wget https://raw.githubusercontent.com/DefTruth/lite.ai.toolkit/main/examples/lite/resources/test_lite_face_detector_3.jpg
+wget https://bj.bcebos.com/paddlehub/fastdeploy/blzeface-1000e.tgz
+
+#使用blazeface-1000e模型
+# CPU推理
+./infer_demo blazeface-1000e/ test_lite_face_detector_3.jpg 0
+# GPU推理
+./infer_demo blazeface-1000e/ test_lite_face_detector_3.jpg 1
+
+运行完成可视化结果如下图所示
+
+<img width="640" src="https://user-images.githubusercontent.com/49013063/206170111-843febb6-67d6-4c46-a121-d87d003bba21.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/cn/faq/use_sdk_on_windows.md)
+
+## BlazeFace C++接口
+
+### BlazeFace类
+
+```c++
+fastdeploy::vision::facedet::BlazeFace(
+        const string& model_file,
+        const string& params_file = "",
+        const string& config_file = "",
+        const RuntimeOption& runtime_option = RuntimeOption(),
+        const ModelFormat& model_format = ModelFormat::PADDLE)
+```
+
+BlazeFace模型加载和初始化，其中model_file为导出的PADDLE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
+> * **config_file**(str): 配置文件路径，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PADDLE格式
+
+#### Predict函数
+
+> ```c++
+> BlazeFace::Predict(cv::Mat& im, FaceDetectionResult* result)
+> ```
+>
+> 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, FaceDetectionResult说明参考[视觉模型预测结果](../../../../../docs/api/vision_results/)
+
+- [模型介绍](../../)
+- [Python部署](../python)
+- [视觉模型预测结果](../../../../../docs/api/vision_results/)
diff --git a/examples/vision/facedet/blazeface/cpp/infer.cc b/examples/vision/facedet/blazeface/cpp/infer.cc
@@ -0,0 +1,94 @@
+// 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_dir, const std::string& image_file) {
+  auto model_file = model_dir + sep + "model.pdmodel";
+  auto params_file = model_dir + sep + "model.pdiparams";
+  auto config_file = model_dir + sep + "infer_cfg.yml";
+  auto option = fastdeploy::RuntimeOption();
+  option.UseCpu();
+  auto model = fastdeploy::vision::facedet::BlazeFace(
+                model_file, params_file, config_file, option);
+  if (!model.Initialized()) {
+    std::cerr << "Failed to initialize." << std::endl;
+    return;
+  }
+
+  auto im = cv::imread(image_file);
+
+  fastdeploy::vision::FaceDetectionResult res;
+  if (!model.Predict(im, &res)) {
+    std::cerr << "Failed to predict." << std::endl;
+    return;
+  }
+  std::cout << res.Str() << std::endl;
+
+  auto vis_im = fastdeploy::vision::VisFaceDetection(im, res);
+  cv::imwrite("vis_result.jpg", vis_im);
+  std::cout << "Visualized result saved in ./vis_result.jpg" << std::endl;
+}
+
+void GpuInfer(const std::string& model_dir, const std::string& image_file) {
+  auto model_file = model_dir + sep + "model.pdmodel";
+  auto params_file = model_dir + sep + "model.pdiparams";
+  auto config_file = model_dir + sep + "infer_cfg.yml";
+  auto option = fastdeploy::RuntimeOption();
+  option.UseGpu();
+  auto model = fastdeploy::vision::facedet::BlazeFace(
+                model_file, params_file, config_file, option);
+  if (!model.Initialized()) {
+    std::cerr << "Failed to initialize." << std::endl;
+    return;
+  }
+
+  auto im = cv::imread(image_file);
+
+  fastdeploy::vision::FaceDetectionResult res;
+  if (!model.Predict(im, &res)) {
+    std::cerr << "Failed to predict." << std::endl;
+    return;
+  }
+  std::cout << res.Str() << std::endl;
+
+  auto vis_im = fastdeploy::vision::VisFaceDetection(im, res);
+  cv::imwrite("vis_result.jpg", vis_im);
+  std::cout << "Visualized result saved in ./vis_result.jpg" << 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 yolov5s-face.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]);
+  }
+  return 0;
+}