确认开发环境已准备FastDeploy C++部署库,参考FastDeploy安装安装预编译的FastDeploy,或根据自己需求进行编译安装。
本文档以PaddleDetection目标检测模型PPYOLOE为例展示CPU上的推理示例
wget https://bj.bcebos.com/paddlehub/fastdeploy/ppyoloe_crn_l_300e_coco.tgz
wget https://bj.bcebos.com/fastdeploy/tests/test_det.jpg
tar xvf ppyoloe_crn_l_300e_coco.tgz
如下C++代码保存为infer_demo.cc
#include "fastdeploy/vision.h"
int main() {
std::string model_file = "ppyoloe_crn_l_300e_coco/model.pdmodel";
std::string params_file = "ppyoloe_crn_l_300e_coco/model.pdiparams";
std::string infer_cfg_file = "ppyoloe_crn_l_300e_coco/infer_cfg.yml";
// 模型推理的配置信息
fastdeploy::RuntimeOption option;
auto model = fastdeploy::vision::detection::PPYOLOE(model_file, params_file, infer_cfg_file, option);
assert(model.Initialized()); // 判断模型是否初始化成功
cv::Mat im = cv::imread("test_det.jpg");
fastdeploy::vision::DetectionResult result;
assert(model.Predict(&im, &result)); // 判断是否预测成功
std::cout << result.Str() << std::endl;
cv::Mat vis_im = fastdeploy::vision::Visualize::VisDetection(im, result, 0.5);
// 可视化结果保存到本地
cv::imwrite("vis_result.jpg", vis_im);
std::cout << "Visualized result save in vis_result.jpg" << std::endl;
return 0;
}
FastDeploy中包含多个依赖库,直接采用g++
或编译器编译较为繁杂,推荐使用cmake进行编译配置。示例配置如下,
假设下载或准备的FastDeploy C++ SDK在/Paddle/Download
目录下,且目录名为fastdeploy_cpp_sdk
,在开发者的项目中只需添加如下代码,即可引入FASTDEPLOY_INCS
和FASTDEPLOY_LIBS
两个变量,分别表示依赖的头文件和库文件
include(/Paddle/Download/fastdeploy_cpp_sdk/FastDeploy.cmake)
PROJECT(infer_demo C CXX)
CMAKE_MINIMUM_REQUIRED (VERSION 3.10)
include(/Path/to/fastdeploy_cpp_sdk/FastDeploy.cmake)
# 添加FastDeploy依赖头文件
include_directories(${FASTDEPLOY_INCS})
add_executable(infer_demo ${PROJECT_SOURCE_DIR}/infer_demo.cc)
target_link_libraries(infer_demo ${FASTDEPLOY_LIBS})
假设当前目录已经准备好infer_demo.cc
和CMakeLists.txt
两个文件,目录结构如下所示,即可进行编译
打开命令行终端,进入infer_demo.cc
和CmakeLists.txt
所在的目录,执行如下命令
mkdir build & cd build
cmake ..
make -j
编译完成后,使用如下命令执行可得到预测结果
./infer_demo
执行时如提示error while loading shared libraries: libxxx.so: cannot open shared object file: No such file...
,说明程序执行时没有找到FastDeploy的库路径,可通过执行如下命令,将FastDeploy的库路径添加到环境变量之后,重新执行二进制程序。
source /Path/to/fastdeploy_cpp_sdk/fastdeploy_init.sh
执行完屏幕会输出如下日志
DetectionResult: [xmin, ymin, xmax, ymax, score, label_id]
415.047180,89.311569, 506.009613, 283.863098, 0.950423, 0
163.665710,81.914932, 198.585342, 166.760895, 0.896433, 0
581.788635,113.027618, 612.623474, 198.521713, 0.842596, 0
267.217224,89.777306, 298.796051, 169.361526, 0.837951, 0
...
...
同时可视化的检测结果图片保存在本地vis_result.jpg
,查看效果如下
上面的编译过程适用于Linux/Mac,Windows上编译流程如下
在Windows菜单中,找到x64 Native Tools Command Prompt for VS 2019
打开,进入infer_demo.cc
和CMakeLists.txt
所在目录,执行如下命令
mkdir build
cd build
cmake .. -G "Visual Studio 16 2019" -A x64
msbuild infer_demo.sln /m /p:Configuration=Release /p:Platform=x64
执行完后,即会在build/Release
目录下生成infer_demo.exe
程序,使用如下命令执行可得到预测结果
Release\infer_demo.exe
Windows上更多SDK使用方式参阅Windows平台使用FastDeploy C++ SDK