Question

我正在尝试迭代许多图像，并将像素值与相应的地面实况图像进行比较。然而，当这样做时，我的计算机的内存使用量增加得非常快，最终程序崩溃了，我得到了这个错误：

OpenCV Error: Insufficient memory (Failed to allocate 259200 bytes) in unknown function, file C:\OpenCV2.3\modules\core\src\alloc.cpp, line 52

我尝试的主要方法如下。我错过了导致这种情况发生的事情吗？有没有更有效的方法呢？

方法1：

void calculateMetrics(cv::Mat gt_frame, cv::Mat result_frame){

    rows = result_frame.rows;
    cols = result_frame.cols;

    int gt;
    int res;

    for(i = 0; i < rows; i++){
        for(j = 0; j < cols; j++){

            gt = gt_frame.at<uchar>(i, j);
            res = result_frame.at<uchar>(i, j);

            if((gt == 255) && (res == 255))     // True positive
                tp++;
            else if((gt == 0) && (res == 0))    // True negative
                tn++;
            else if((gt == 0) && (res == 255))  // False positive
                fp++;
            else if((gt == 255) && (res == 0))  // False negative
                fn++;
        }
    }
}

方法2：

void calculateMetrics(cv::Mat gt_frame, cv::Mat result_frame){

    rows = result_frame.rows;
    cols = result_frame.cols;

    uchar *gt;
    uchar *res;

    for(i = 0; i < rows; ++i){
        gt = gt_frame.ptr(i);
        res = result_frame.ptr(i);

        for(j = 0; j < cols; ++j){
            *gt++;
            *res++;

            if((*gt == 255) && (*res == 255))       // True positive
                tp++;
            else if((*gt == 0) && (*res == 0))  // True negative
                tn++;
            else if((*gt == 0) && (*res == 255))    // False positive
                fp++;
            else if((*gt == 255) && (*res == 0))    // False negative
                fn++;
       }
    }
}

致电代码：

void EvaluationFramework::processResults(int video_num){
    num_frames = getNumFrames(video_num);

    for(j = 0; j < num_frames; j++){

        gt_frame = getGroundTruthFrame(video_num, j);
        result_frame = getResultFrame(video_num, j);

        calculateMetrics(gt_frame, result_frame);
    }

    std::cout << "TP: " << tp << std::endl;
    std::cout << "TN: " << tn << std::endl;
    std::cout << "FP: " << fp << std::endl;
    std::cout << "FN: " << fn << std::endl;

    precision = getPrecision((double)tp, (double)fp);
    recall = getRecall((double)tp, (double)fn);
    fmeasure = getFmeasure(precision, recall);

    std::cout << "Precision: " << precision << std::endl;
    std::cout << "Recall: " << recall << std::endl;
    std::cout << "F-Measure: " << fmeasure << std::endl;

    insertResults(0, video_num, precision, recall, fmeasure);
}

getGroundTruthFrame和getResultFrame函数只是从文件夹读取相关图像，getPrecision，getRecall和getFmeasure执行相当简单的计算，insertResults将计算值插入数据库。

有效地迭代cvMat图像

0 个答案: