Zi 字媒體

fanfuhan OpenCV 教學100 ~ opencv-100-HOG特徵與行人 檢測/定位/標記 資料來源: https://fanfuhan.github.io/ https://fanfuhan.github.io/2019/05/16/opencv-100/ GITHUB:https://github.com/jash-git/fanfuhan_ML_OpenCV HOG（定向直方圖）特徵在對象識別與模式匹配中是一種常見的特徵提取算法，是基於本地目標塊進行特徵直方圖提取的一種算法，對象局部的變形與光照影響有很好的穩定性，最初是用HOG特徵來來識別人像，通過HOG特徵提取+ SVM訓練，可以得到很好的效果，OpenCV已經有了。 C++ #include #include using namespace cv; using namespace std; int main(int argc, char** argv) { Mat src = imread("D:/images/pedestrian.png"); if (src.empty()) { printf("could not load image..\n"); return -1; } imshow("input", src); HOGDescriptor *hog = new HOGDescriptor(); hog->setSVMDetector(hog->getDefaultPeopleDetector()); vector objects; hog->detectMultiScale(src, objects, 0.0, Size(4, 4), Size(8, 8), 1.25); for (int i = 0; i < objects.size(); i++) { rectangle(src, objects[i], Scalar(0, 0, 255), 2, 8, 0); } imshow("result", src); waitKey(0); return 0; } Python """ HOG特征与行人检测 """ import cv2 as cv src = cv.imread("images/pedestrian.png") hog = cv.HOGDescriptor() hog.setSVMDetector(cv.HOGDescriptor_getDefaultPeopleDetector()) # detect people in image (rects, weights) = hog.detectMultiScale(src, winStride=(4, 4), padding=(8, 8), scale=1.25, useMeanshiftGrouping=False) for (x, y, w, h) in rects: cv.rectangle(src, (x, y), (x + w, y + h), (0, 255, 0), 2) cv.imshow("hog-detector", src) cv.waitKey(0) cv.destroyAllWindows()