Ubuntu20.04下Qt配置OpenCV和libtorch

whut_52xj

已于 2023-02-28 21:49:50 修改

阅读量751

点赞数

分类专栏： libtorch学习文章标签： qt opencv 开发语言

于 2023-02-28 21:24:29 首次发布

本文链接：https://blog.csdn.net/weixin_48192326/article/details/129270216

版权

libtorch学习专栏收录该内容

4 篇文章 0 订阅

订阅专栏

Ubuntu20.04下Qt配置OpenCV和libtorch

Qt版本：5.14.2
OpenCV：4.5.1
libtorch：1.2.1

1.OpenCV配置

在.pro项目文件中添加如下内容

INCLUDEPATH += /usr/local/include \
               /usr/local/include/opencv4 \
               /usr/local/include/opencv4/opencv2

LIBS += /usr/local/lib/libopencv*

OpenCV测试代码及结果

void opencvandlibtorch_test::opencv_test(){
    cout <<"opencv test" << endl;
    cout <<"opencv version: " << CV_VERSION << endl;

    Mat src = imread("/home/zxm/图片/opencv.png");

    if(src.empty()){
        cout << "could not load image" << endl;
        return;
    }

    namedWindow("src",WINDOW_NORMAL);

    imshow("src",src);
    waitKey(0);
}

在这里插入图片描述

2.libtorch配置

在.pro项目文件中添加如下内容

INCLUDEPATH += /home/zxm/libtorch/include
INCLUDEPATH += /home/zxm/libtorch/include/torch/csrc/api/include

#.so
LIBS += -L/home/zxm/libtorch/lib\
        -lc10\
        -lshm\
        -lc10d_cuda_test\
        -ltorchbind_test\
        -lnnapi_backend\
        -lc10_cuda\
        -ltorch_python\
        -ltorch_global_deps\
        -ljitbackend_test\
        -ltorch_cuda\
        -lcaffe2_nvrtc\
        -lbackend_with_compiler\
        -ltorch\
        -ltorch_cuda_linalg\
        -ltorch_cuda_cpp\
        -ltorch_cuda_cu\
        -ltorch_cpu\
        -INCLUDE:?warp_size@cuda@at@@YAHXZ\
        -Wl,--no-as-needed -ltorch_cuda #force to link torch_cuda
        
QMAKE_LFLAGS += -INCLUDE:?searchsorted_cuda@native@at@@YA?AVTensor@2@AEBV32@0_N1@Z
CONFIG += c++14

libtorch测试代码及结果

void opencvandlibtorch_test::libtorch_test(){
    cout << "libtorch test " << endl;
    cout << "LibTorch version : "
              << TORCH_VERSION_MAJOR << "."
              << TORCH_VERSION_MINOR << "."
              << TORCH_VERSION_PATCH << endl;
    cout << "PyTorch version: " << TORCH_VERSION <<endl;


    torch::Tensor tensor = torch::rand({2, 3});
    std::cout << tensor << std::endl;


   //测试是否能使用GPU
    if(torch::cuda::is_available()){
        cout << "can use GPU" << endl;
    }
    else{
        cout << "can not use GPU" << endl;
    }
}