如何使用 opencv 加載 darknet yolo 預訓練模型？

如何使用 opencv 加載 darknet yolo 預訓練模型？

opencv 版本 > 3.4 以上

constexpr const char *image_path = "darknet.jpg";//待檢測圖片
constexpr const char *darknet_cfg = "darknet.cfg";//網絡文件
constexpr const char *darknet_weights = "darknet.weights";//訓練模型
const std::vector<std::string> class_labels = {"darknet","yolo"};//類標籤

void darknetDetection(const std::string &path,const std::string &darknet_cfg,const std::string &darknet_weights,std::vector<std::string> class_labels,float confidenceThreshold)
{
    // 加載模型
    cv::dnn::Net net = cv::dnn::readNetFromDarknet(darknet_cfg,darknet_weights);

    // 加載標籤集
    std::vector<std::string> classLabels = class_labels;

    // 讀取待檢測圖片
    cv::Mat img = cv::imread(path);
    cv::Mat blob = cv::dnn::blobFromImage(img,1.0/255.0,{416,416},0.00392,true);
    net.setInput(blob,"data");

    // 檢測
    cv::Mat detectionMat = net.forward("detection_out");// 6 845 1 W x H x C

    // 獲取網絡每層的用時並獲取總用時
    std::vector<double> layersTimings;
    double freq = cv::getTickFrequency() / 1000;
    double time = net.getPerfProfile(layersTimings) / freq;
    std::ostringstream ss;
    ss << "detection time: " << time << " ms";
    // 繪製總用時至原始圖片
    cv::putText(img, ss.str(), cv::Point(20, 20), 0, 0.5, cv::Scalar(0, 0, 255));

    // 遍歷全部結果集
    for(int i = 0;i < detectionMat.rows;++i){
        const int probability_index = 5;
        const int probability_size = detectionMat.cols - probability_index;
        float *prob_array_ptr = &detectionMat.at<float>(i, probability_index);
        size_t objectClass = std::max_element(prob_array_ptr, prob_array_ptr + probability_size) - prob_array_ptr;
        float confidence = detectionMat.at<float>(i, (int)objectClass + probability_index);

        // 比較置信度並繪製知足條件的置信度
        if (confidence > confidenceThreshold)
        {
            float x = detectionMat.at<float>(i, 0);
            float y = detectionMat.at<float>(i, 1);
            float width = detectionMat.at<float>(i, 2);
            float height = detectionMat.at<float>(i, 3);

            int xLeftBottom = static_cast<int>((x - width / 2) * img.cols);
            int yLeftBottom = static_cast<int>((y - height / 2) * img.rows);
            int xRightTop = static_cast<int>((x + width / 2) * img.cols);
            int yRightTop = static_cast<int>((y + height / 2) * img.rows);

            cv::Rect object(xLeftBottom, yLeftBottom,xRightTop - xLeftBottom,yRightTop - yLeftBottom);//x y w h
            cv::rectangle(img, object, cv::Scalar(0, 0, 255), 2, 8);

            // 判斷類id是否符合標籤範圍並繪製該標籤，也就是矩陣的下標索引
            if (objectClass < classLabels.size())
            {
                cv::String label = cv::String(classLabels[objectClass]) + ": " + std::to_string(confidence);
                int baseLine = 0;
                cv::Size labelSize = cv::getTextSize(label,cv::FONT_HERSHEY_SIMPLEX, 0.5, 1, &baseLine);
                cv::rectangle(img, cv::Rect(cv::Point(xLeftBottom, yLeftBottom),cv::Size(labelSize.width, labelSize.height + baseLine)),cv::Scalar(255, 255, 255), cv::FILLED);
                cv::putText(img, label, cv::Point(xLeftBottom, yLeftBottom + labelSize.height),cv::FONT_HERSHEY_SIMPLEX, 0.5, cv::Scalar(0, 0, 0));
            }
        }
    }

    // 顯示圖片
    cv::imshow("Darknet",img);
    cv::waitKey(0);
}