热门标签
热门文章
- 1Git 中文安装教程_git 安装默认了中文安装路径怎么处理
- 2Git下载与安装
- 3NLP学习（二）—中文分词技术_正向最大匹配算法流程图
- 4使用jedis连接云服务器搭建的redis集群循环抛出connection timed out的问题解决_io.lettuce.core.cluster.topology.clustertopologyre
- 5android自定义application,安卓以widget方式集成，不能使用自定义的application，会报错，如何解决...
- 6山东大学 电子商务_多媒体内容与网络出版层
- 7我的软件测试面试经历，7轮高强度面试顺利入职_软件测试怎么面试高级别
- 8一文搞懂：什么是数字化IT运维？（三）
- 9【Python 函数和面向对象的实践应用】合肥工业大学python实验二题解_使用freqdict = eval(input()) 读入单词词频字典,再读入一段英文,默认按照英文
- 10写给开发者的软件架构实战：如何实施持续集成和持续部署
当前位置: article > 正文
windows下，opencv4.8.0+cuda编译及在vs2022中使用的详细教程_vs2022打包cuda程序
作者：weixin_40725706 | 2024-07-17 12:18:34
赞
踩
vs2022打包cuda程序
1、准备好opencv、opencv_contrib-4.8.0、cmake、vs2022，cuda
配置不好的话，可直接找我要配置好的build文件：可进群：482742815，也可加qq：1401682937，请备注好，否则不同意
另外，我上传了一部分文件，可以查看链接：https://download.csdn.net/download/Word_And_Me_/88605111
opencv：Release OpenCV 4.8.0 · opencv/opencv · GitHub
opencv：Release OpenCV 4.8.0 · opencv/opencv · GitHub
opencv_contrib：Release 4.8.0 · opencv/opencv_contrib · GitHub
cmake：https://cmake.org/files/v3.23/cmake-3.23.0-rc1-windows-i386.msi
cuda我使用的是11.7，可以配置其他版本一样，具体不再介绍
2、开始使用cmake编译
首先，选择好路径，source为本来就有的，cuda_build为自定义名称，设置好点击configure
点击configure之后会弹出选择编译器，选好之后finish
编译过程中，或出现一些错误（红字体），这里第一个关于python的不用理会
另外一些问题是因为一些包不能下载，例如ffmpeg，ippicv
对应的文件可以在source源码中，找到对应的cmake文件，里面有下载地址，如果实在下不下来，我后面将提供文件
文件下载的地址为：（注意：新下载的文件命名需要在前面再加一段哈希字符，可在上面所说的cmake文件中找到）
cmake中需要查找的信息：
缺失的包下载完毕，并存储到相应位置后，点击generate
generate之后，配置opencv_ex目录，如下图：
搜索cuda，全部勾选，如图：
搜索test，取消勾选前三个，没有用，如图：
完成以上内容后，继续点击configure，如图：（注意，这一次的configure时间稍微较长）
configure完成之后，搜索cuda，会发现有一堆东西，这里我们找到cuda_fast_math勾选，其余不要动，如图：
在上个configure之后，你还会发现一些错误，这里也是因为下载失败，需要手动下载：（这里我选择不要face，其余两个尝试下载，同样方法找到cmake文件，找url，这里的cmake文件在opencv_contrib/moudules中）
这里如果你不用xfeatures2d，wechat_qrcode，也可以搜索相应的文件，取消勾选；以上操作完成之后，点击gengrate，如下图：
generate完成之后，进入对应的文件夹（cuda_build），打开opencv.sln
将debug改为release，生成对应的文件，如图：（生成一般需要1-2个小时，我使用了3个半小时，这里根据自己的设备不同，时间不一样，我了解到有人使用8个小时）
all_build生成完毕：
install生成完毕：
使用配置
1、打开自己创建的项目，右键-->属性-->vc++目录
包含目录：
库目录：
2、打开自己创建的项目，右键-->属性-->链接器-->输入-->编辑附加依赖项
这里依赖项可能是一堆，没有集成起来，可以使用Python获得所有的.lib文件,如果想要一个文件可以在编译阶段勾选，然后生成即可获得一个封装好的，但是存在风险，一旦有一个编译不成功，则不会生成这一个集成的文件，勾选项如下图：
我的有.lib文件有：
提供一个代码：
- import os
-
- path = os.getcwd()
- for file in os.listdir(path):
- with open('all.txt', 'a', encoding='utf-8') as f:
- if file.endswith('.lib'):
- newpath = path + "/" + file
- newpath = newpath.replace("\\", "/")
- print(newpath)
- f.write(f"{newpath}\n")
我目前完成的测试：
- #include <iostream>
- #include <opencv2/opencv.hpp>
- #include <opencv2/core/cuda.hpp>
- #include <opencv2/cudaimgproc.hpp> // cvtColor
- #include <opencv2/cudafilters.hpp> // createGaussianFilter()
- #include <opencv2/cudafeatures2d.hpp>
- #include <opencv2/xfeatures2d.hpp>
- #include <opencv2/xfeatures2d/cuda.hpp>
- #include <opencv2/cudaimgproc.hpp>
- #include <opencv2/cudaarithm.hpp>
- #include <vector>
-
-
-
- int main() {
- // =================================cuda 测试=======================================
- /*cv::cuda::printCudaDeviceInfo(cv::cuda::getDevice());
- int count = cv::cuda::getCudaEnabledDeviceCount();
- std::cout << "可以使用的数量：" << count << std::endl;
- return 0;*/
- // ==================================灰度转换=======================================
- /*cv::Mat src_host = cv::imread("C:/Users/admin/Desktop/lena.jpg");
- cv::cuda::GpuMat src, gray;
- src.upload(src_host);
- cv::cuda::cvtColor(src, gray, cv::COLOR_BGR2GRAY);
- cv::Mat gray_host;
- gray.download(gray_host);
- cv::imshow("src", src_host);
- cv::imshow("gray", gray_host);
- cv::waitKey();
- cv::destroyAllWindows();
- return 0;*/
- // ==================================高斯模糊=======================================
- /*cv::Mat h_img1 = cv::imread("C:/Users/admin/Desktop/lenasp.jpg");
- cv::cuda::GpuMat d_img1, d_result3x3, d_result5x5, d_result7x7;
- // 加载数据
- d_img1.upload(h_img1);
- // 创建高斯
- auto filter3x3 = cv::cuda::createGaussianFilter(CV_8UC3, CV_8UC3, cv::Size(3, 3), 5);
- auto filter5x5 = cv::cuda::createGaussianFilter(CV_8UC3, CV_8UC3, cv::Size(5, 5), 5);
- auto filter7x7 = cv::cuda::createGaussianFilter(CV_8UC3, CV_8UC3, cv::Size(7, 7), 5);
- // 执行
- filter3x3->apply(d_img1, d_result3x3);
- filter5x5->apply(d_img1, d_result5x5);
- filter7x7->apply(d_img1, d_result7x7);
- // 获取结果
- cv::Mat h_result3x3, h_result5x5, h_result7x7;
- d_result3x3.download(h_result3x3);
- d_result5x5.download(h_result5x5);
- d_result7x7.download(h_result7x7);
- // 显示
- cv::imshow("original image", h_img1);
- cv::imshow("Blurred with kernel size 3x3", h_result3x3);
- cv::imshow("Blurred with kernel size 5x5", h_result5x5);
- cv::imshow("Blurred with kernel size 7x7", h_result7x7);
- cv::waitKey();
- cv::destroyAllWindows();
- return 0;*/
- // ==============================shi-tomas角点检测============================
- /*cv::Mat src_host = cv::imread("C:/Users/admin/Desktop/jianzhu.jpg");
- cv::imshow("input", src_host);
- cv::cuda::GpuMat src, gray, corners;
- cv::Mat dst;
- src.upload(src_host);
- cv::cuda::cvtColor(src, gray, cv::COLOR_BGR2GRAY);
- auto corner_detector = cv::cuda::createGoodFeaturesToTrackDetector(gray.type(), 1000, 0.01, 15, 3);
- corner_detector->detect(gray, corners);
- corners.download(dst);
- printf("detected corners %d....\n", corners.cols);
- printf("detected corners %d....\n", dst.cols);
- cv::RNG rng(123456);
- for (int i = 0; i < dst.cols; i++) {
- int r = rng.uniform(0, 255);
- int g = rng.uniform(0, 255);
- int b = rng.uniform(0, 255);
- cv::Point2f pt = dst.at<cv::Point2f>(0, i);
- circle(src_host, pt, 3, cv::Scalar(b, g, r), 2, 8, 0);
- }
- imshow("corner detect", src_host);
- cv::waitKey();
- cv::destroyAllWindows();
- return 0;*/
-
- // ================================双边模糊（美颜）===========================
- /*try {
- cv::Mat src_host = cv::imread("C:/Users/admin/Desktop/hh.jpg");
- cv::imshow("input", src_host);
- cv::cuda::GpuMat src(src_host);
- cv::cuda::GpuMat dst;
- cv::cuda::bilateralFilter(src, dst, 0, 100, 15, 4);
- cv::Mat dst_host;
- dst.download(dst_host);
- cv::imshow("result", dst_host);
- }
- catch (const cv::Exception& ec) {
- std::cout << "Error" << ec.what() << std::endl;
- }
- cv::waitKey();
- return 0;*/
-
- // ==============================ORB 特征匹配==========================
- /*// cpu data
- cv::Mat h_object_image = cv::imread("C:/Users/admin/Desktop/jianzhu1.jpg", 0);
- cv::Mat h_scene_image = cv::imread("C:/Users/admin/Desktop/jianzhu.jpg", 0);
-
- assert(h_object_image != nullptr && h_scene_image != nullptr), "======没有读取成功=====";
- cv::waitKey();
- // gpu data
- cv::cuda::GpuMat d_object_image;
- cv::cuda::GpuMat d_scene_image;
-
- cv::cuda::GpuMat d_keypoints_scene, d_keypoints_object; // GPU key points
- std::vector<cv::KeyPoint> h_keypoints_scene, h_keypoints_object; // CPU key points
- cv::cuda::GpuMat d_descriptors_scene, d_descriptors_object; // GPU descriptor
-
- // Image CPU uploaded to GPU
- d_object_image.upload(h_object_image);
- d_scene_image.upload(h_scene_image);
-
- // 对象检测
- auto orb = cv::cuda::ORB::create();
- // 检测特征点并提取相应的描述符
- // Detect feature points and extract corresponding descriptors
- orb->detectAndCompute(d_object_image, cv::cuda::GpuMat(), h_keypoints_object, d_descriptors_object);
- orb->detectAndCompute(d_scene_image, cv::cuda::GpuMat(), h_keypoints_scene, d_descriptors_scene);
-
- // 暴力匹配器
- // Brute Force Violence Matcher
- cv::Ptr<cv::cuda::DescriptorMatcher> matcher = cv::cuda::DescriptorMatcher::createBFMatcher(cv::NORM_HAMMING);
- std::vector<std::vector<cv::DMatch> > d_matches;
- matcher->knnMatch(d_descriptors_object, d_descriptors_scene, d_matches, 2);
-
- std::cout << "match size:" << d_matches.size() << std::endl;
- std::vector< cv::DMatch > good_matches;
- for (int k = 0; k < std::min(h_keypoints_object.size() - 1, d_matches.size()); k++)
- {
- if ((d_matches[k][0].distance < 0.9 * (d_matches[k][1].distance)) &&
- ((int)d_matches[k].size() <= 2 && (int)d_matches[k].size() > 0))
- {
- good_matches.push_back(d_matches[k][0]);
- }
- }
- std::cout << "size:" << good_matches.size() << std::endl;
- cv::Mat h_image_result;
- cv::drawMatches(h_object_image, h_keypoints_object, h_scene_image, h_keypoints_scene,
- good_matches, h_image_result, cv::Scalar::all(-1), cv::Scalar::all(-1),
- std::vector<char>(), cv::DrawMatchesFlags::DEFAULT);
- std::vector<cv::Point2f> object, scene;
- for (int i = 0; i < good_matches.size(); i++) {
- object.push_back(h_keypoints_object[good_matches[i].queryIdx].pt);
- scene.push_back(h_keypoints_scene[good_matches[i].trainIdx].pt);
- }
- cv::Mat Homo = cv::findHomography(object, scene, cv::RANSAC);
- std::vector<cv::Point2f> corners(4), scene_corners(4);
- corners[0] = cv::Point(0, 0);
- corners[1] = cv::Point(h_object_image.cols, 0);
- corners[2] = cv::Point(h_object_image.cols, h_object_image.rows);
- corners[3] = cv::Point(0, h_object_image.rows);
- cv::perspectiveTransform(corners, scene_corners, Homo);
-
- cv::line(h_image_result, scene_corners[0] + cv::Point2f(h_object_image.cols, 0), scene_corners[1] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::line(h_image_result, scene_corners[1] + cv::Point2f(h_object_image.cols, 0), scene_corners[2] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::line(h_image_result, scene_corners[2] + cv::Point2f(h_object_image.cols, 0), scene_corners[3] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::line(h_image_result, scene_corners[3] + cv::Point2f(h_object_image.cols, 0), scene_corners[0] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::imshow("Good Matches & Object detection", h_image_result);
- cv::waitKey(0);
-
- return 0;*/
- // ==============================surf 特征检测==============================
- cv::Mat h_object_image = cv::imread("C:/Users/admin/Desktop/jianzhu1.jpg", 0);
- cv::Mat h_scene_image = cv::imread("C:/Users/admin/Desktop/jianzhu.jpg", 0);
-
- cv::cuda::GpuMat d_object_image;
- cv::cuda::GpuMat d_scene_image;
-
- cv::cuda::GpuMat d_keypoints_scene, d_keypoints_object;
- std::vector<cv::KeyPoint> h_keypoints_scene, h_keypoints_object;
- cv::cuda::GpuMat d_descriptors_scene, d_descriptors_object;
-
- d_object_image.upload(h_object_image);
- d_scene_image.upload(h_scene_image);
-
- // 对象检测
- cv::cuda::SURF_CUDA surf(100);
- // 检测特征点并提取相应的描述符
- surf(d_object_image, cv::cuda::GpuMat(), d_keypoints_object, d_descriptors_object);
- surf(d_scene_image, cv::cuda::GpuMat(), d_keypoints_scene, d_descriptors_scene);
-
- // 暴力匹配器
- cv::Ptr<cv::cuda::DescriptorMatcher> matcher = cv::cuda::DescriptorMatcher::createBFMatcher();
- std::vector<std::vector<cv::DMatch>> d_matches;
- matcher->knnMatch(d_descriptors_object, d_descriptors_scene, d_matches, 2);
-
- surf.downloadKeypoints(d_keypoints_scene, h_keypoints_scene);
- surf.downloadKeypoints(d_keypoints_object, h_keypoints_object);
-
- std::vector<cv::DMatch> good_matches;
- for (int k = 0; k < std::min(h_keypoints_object.size() - 1, d_matches.size()); k++)
- {
- if ((d_matches[k][0].distance < 0.6 * (d_matches[k][1].distance)) && ((int)d_matches[k].size() <= 2 && (int)d_matches[k].size() > 0))
- {
- good_matches.push_back(d_matches[k][0]);
- }
- }
- std::cout << "size:" << good_matches.size();
- cv::Mat h_image_result;
- cv::drawMatches(h_object_image, h_keypoints_object, h_scene_image, h_keypoints_scene,
- good_matches, h_image_result, cv::Scalar::all(-1), cv::Scalar::all(-1),
- std::vector<char>(), cv::DrawMatchesFlags::DEFAULT);
-
- std::vector<cv::Point2f> object;
- std::vector<cv::Point2f> scene;
- for (int i = 0; i < good_matches.size(); i++)
- {
- object.push_back(h_keypoints_object[good_matches[i].queryIdx].pt);
- scene.push_back(h_keypoints_scene[good_matches[i].trainIdx].pt);
- }
- cv::Mat Homo = cv::findHomography(object, scene, cv::RANSAC);
- std::vector<cv::Point2f> corners(4);
- std::vector<cv::Point2f> scene_corners(4);
- corners[0] = cv::Point(0, 0);
- corners[1] = cv::Point(h_object_image.cols, 0);
- corners[2] = cv::Point(h_object_image.cols, h_object_image.rows);
- corners[3] = cv::Point(0, h_object_image.rows);
- cv::perspectiveTransform(corners, scene_corners, Homo);
-
- cv::line(h_image_result, scene_corners[0] + cv::Point2f(h_object_image.cols, 0), scene_corners[1] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::line(h_image_result, scene_corners[1] + cv::Point2f(h_object_image.cols, 0), scene_corners[2] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::line(h_image_result, scene_corners[2] + cv::Point2f(h_object_image.cols, 0), scene_corners[3] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::line(h_image_result, scene_corners[3] + cv::Point2f(h_object_image.cols, 0), scene_corners[0] + cv::Point2f(h_object_image.cols, 0), cv::Scalar(255, 0, 0), 4);
- cv::imshow("Good Matches & Object detection", h_image_result);
- cv::waitKey(0);
- return 0;
-
- // =================================ORB特征检测================================
- /*cv::Mat h_image = cv::imread("C:/Users/admin/Desktop/jianzhu.jpg", 0);
- cv::Ptr<cv::cuda::ORB> detector = cv::cuda::ORB::create();
- std::vector<cv::KeyPoint> keypoints;
- cv::cuda::GpuMat d_image;
- d_image.upload(h_image);
- detector->detect(d_image, keypoints);
- cv::drawKeypoints(h_image, keypoints, h_image);
- cv::imshow("Final Result", h_image);
- cv::waitKey(0);
- return 0;*/
- }
-
声明：本文内容由网友自发贡献，不代表【wpsshop博客】立场，版权归原作者所有，本站不承担相应法律责任。如您发现有侵权的内容，请联系我们。转载请注明出处：https://www.wpsshop.cn/w/weixin_40725706/article/detail/839860
推荐阅读
相关标签
