C+++libtorch1.7（release版本）+pytorch1.7+cuda10.1+opencv4.7.0实现YOLOv5模型调用

1、libtorch与opencv配置主要参考：
（1）https://zhuanlan.zhihu.com/p/513571175
（2）官网地址：https://opencv.org/releases/blog.csdnimg.cn/cd85285bff894261b8785229107ba99e.png)
下载后安装

配置环境变量

下载Libtorch：Libtorch版本要与pytorch版本严格对应

​
Windows下：
Libtorch 1.0.0
cpu-release: https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.0.0.zip
 
cu100-release: https://download.pytorch.org/libtorch/cu100/libtorch-win-shared-with-deps-1.0.0.zip
 
1.0没有debug的版本，从1.1开始有
 
Libtorch 1.1.0
cpu-debug: https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.1.0.zip
 
cpu-release: https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.1.0.zip
 
cu100-debug: https://download.pytorch.org/libtorch/cu100/libtorch-win-shared-with-deps-debug-1.1.0.zip
 
cu100-release: https://download.pytorch.org/libtorch/cu100/libtorch-win-shared-with-deps-1.1.0.zip
 
Libtorch 1.1.0 —— 1.5.0：  
按照1.1.0的格式来，只需要修改最后的几个数字或对应的CUDA的版本
 
Libtorch 1.6.0
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.6.0%2Bcpu.zip
 
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.6.0%2Bcpu.zip
 
cu101-debug: https://download.pytorch.org/libtorch/cu101/libtorch-win-shared-with-deps-debug-1.6.0%2Bcu101.zip
 
cu101-release: https://download.pytorch.org/libtorch/cu101/libtorch-win-shared-with-deps-1.6.0%2Bcu101.zip
 
cu102-debug: https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-debug-1.6.0.zip
 
cu102-release: https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-1.6.0.zip
 
Libtorch 1.7.0
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.7.0%2Bcpu.zip
 
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.7.0%2Bcpu.zip
 
cu101-debug: https://download.pytorch.org/libtorch/cu101/libtorch-win-shared-with-deps-debug-1.7.0%2Bcu101.zip
 
cu101-release: https://download.pytorch.org/libtorch/cu101/libtorch-win-shared-with-deps-1.7.0%2Bcu101.zip
 
cu102-debug:https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-debug-1.7.0.zip
 
cu102-debug:https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-1.7.0.zip
 
Libtorch 1.8.0
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.8.0%2Bcpu.zip
 
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.8.0%2Bcpu.zip
 
cu102-debug:​​​​​​​ https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-debug-1.8.0.zip
 
cu102-release: https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-1.8.0.zip
 
Libtorch 1.9.0
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.9.0%2Bcpu.zip
 
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.9.0%2Bcpu.zip
 
cu102-debug:https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-debug-1.9.0%2Bcu102.zip
 
cu102-release:https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-1.9.0%2Bcu102.zip
 
libtorch 1.10.0
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.10.0%2Bcpu.zip
 
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.10.0%2Bcpu.zip
 
cu102-debug: https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-debug-1.10.0%2Bcu102.zip
 
cu102-release: https://download.pytorch.org/libtorch/cu102/libtorch-win-shared-with-deps-1.10.0%2Bcu102.zip
 
libtorch 1.11.0
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.11.0%2Bcpu.zip
 
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.11.0%2Bcpu.zip
 
cu113-debug: https://download.pytorch.org/libtorch/cu113/libtorch-win-shared-with-deps-debug-1.11.0%2Bcu113.zip
 
cu113-release: https://download.pytorch.org/libtorch/cu113/libtorch-win-shared-with-deps-1.11.0%2Bcu113.zip
 
libtorch 1.12.0
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.12.0%2Bcpu.zip
 
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.12.0%2Bcpu.zip
 
cu113-debug: https://download.pytorch.org/libtorch/cu113/libtorch-win-shared-with-deps-debug-1.12.0%2Bcu113.zip
 
cu113-release: https://download.pytorch.org/libtorch/cu113/libtorch-win-shared-with-deps-1.12.0%2Bcu113.zip
 
cu116-debug: https://download.pytorch.org/libtorch/cu116/libtorch-win-shared-with-deps-debug-1.12.0%2Bcu116.zip
 
cu116-release: https://download.pytorch.org/libtorch/cu116/libtorch-win-shared-with-deps-1.12.0%2Bcu116.zip
 
libtorch 1.13.0
cpu-release：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-1.13.0%2Bcpu.zip
 
cpu-debug：https://download.pytorch.org/libtorch/cpu/libtorch-win-shared-with-deps-debug-1.13.0%2Bcpu.zip
 
cu116-debug: https://download.pytorch.org/libtorch/cu116/libtorch-win-shared-with-deps-debug-1.13.0%2Bcu116.zip
 
cu116-release: https://download.pytorch.org/libtorch/cu116/libtorch-win-shared-with-deps-1.13.0%2Bcu116.zip
 
cu117-debug: https://download.pytorch.org/libtorch/cu117/libtorch-win-shared-with-deps-debug-1.13.0%2Bcu117.zip
 
cu117-release: https://download.pytorch.org/libtorch/cu117/libtorch-win-shared-with-deps-1.13.0%2Bcu117.zip
 

解压到本地：

2、配置VS项目属性（opencv和libtorch）：
新建一个项目（所建为控制台应用程序）：
项目配置改为 Release x64环境。
打开如下设置：
（1）项目>>>属性>>>VC++目录，分别在包含目录和库目录中添加自己libtorch包的include文件路径和lib文件路径。

（2）项目>>>属性>>>链接器>>>输入，在附加依赖项中添加需要的库文件名称，如果觉得太麻烦，那么可以把库目录中的以.lib为后缀名的文件名全部添加上去。

（3）项目>>>Project2属性>>>C/C++，进行两个改动：第一，“常规”目标栏中的“SDL检查”改为“否”；第二，“语言”目标栏中的“符合模式”改为“否”。否则会报错：“std”: 不明确的符号。

3、测试

使用如下代码：
选择releaseX64

#include <torch/script.h> 
#include <torch/torch.h> 
#include <iostream>
#include <memory>

int main(int argc, const char* argv[]) {
            
	std::cout << "cuda::is_available():" << torch::cuda::is_available() << std::endl;
	torch::DeviceType device_type = at::kCPU; // 定义设备类型
	if (torch::cuda::is_available())
		device_type = at::kCUDA;
}

4、YOLOv5推理：
源码地址：https://github.com/Nebula4869/YOLOv5-LibTorch
下载后解压，并创建build文件夹：

对CMakeLists进行更改

将set(CMAKE_PREFIX_PATH后面路径更改成自己的libtorch路径

在build文件夹中打开cmd，运行：

cmake ..

此时要安装cmake这个软件：https://cmake.org/download/

编译成功界面：

接着在vs2019中打开编译好的build文件中的工程进行执行：


执行成功后可以再下面文件夹生成exe文件：

若执行过程中提示缺少DLL或者lib文件，则可以把libtorch下的相关文件添加在.exe同级文件夹中。

至此完成。之后会讲一下如何对此工程进行打包。