Troubleshooting

• Vehicle detection 모델을 YOLO v3로 바꾼 후 발생하는 문제

  • bash run.sh -i samples/test -o results/yolov3 -c results/yolov3/results.csv

    • OSError: libdarknet.so: cannot open shared object file: No such file or directory

      • darknet/python/darknet.py 파일 내부에 libdarknet.so 위치 변경

      • lib = CDLL(“libdarknet.so”, RTLD_GLOBAL) → lib = CDLL(“darknet/libdarknet.so”, RTLD_GLOBAL)

    • ValueError: need more than 0 values to unpack

      • 에러 내용

      Traceback (most recent call last):
      File "vehicle-detection.py", line 45, in <module>
          R,_ = detect(vehicle_net, vehicle_meta, img_path ,thresh=vehicle_threshold)
      ValueError: need more than 0 values to unpack
      

      • darknet/python/darknet.py 파일의 detect 함수에서 wh 값을 추가하고 반환함

        • Before

        def detect(net, meta, image, thresh=.5, hier_thresh=.5, nms=.45):
            im = load_image(image, 0, 0)
            num = c_int(0)
            pnum = pointer(num)
            predict_image(net, im)
            dets = get_network_boxes(net, im.w, im.h, thresh, hier_thresh, None, 0, pnum)
            num = pnum[0]
            if (nms): do_nms_obj(dets, num, meta.classes, nms);
        
            res = []
            for j in range(num):
                for i in range(meta.classes):
                    if dets[j].prob[i] > 0:
                        b = dets[j].bbox
                        res.append((meta.names[i], dets[j].prob[i], (b.x, b.y, b.w, b.h)))
            res = sorted(res, key=lambda x: -x[1])
            free_image(im)
            free_detections(dets, num)
            return res
        

        • After

        def detect(net, meta, image, thresh=.5, hier_thresh=.5, nms=.45):
            im = load_image(image, 0, 0)
            num = c_int(0)
            pnum = pointer(num)
            predict_image(net, im)
            dets = get_network_boxes(net, im.w, im.h, thresh, hier_thresh, None, 0, pnum)
            num = pnum[0]
            if (nms): do_nms_obj(dets, num, meta.classes, nms);
        
            res = []
            for j in range(num):
                for i in range(meta.classes):
                    if dets[j].prob[i] > 0:
                        b = dets[j].bbox
                        res.append((meta.names[i], dets[j].prob[i], (b.x, b.y, b.w, b.h)))
            res = sorted(res, key=lambda x: -x[1])
            wh = (im.w,im.h)
            free_image(im)
            free_detections(dets, num)
            return res, wh
        

    • CUDA Error: out of memory (Link)

      • 에러 내용

      CUDA Error: out of memory
      python: ./src/cuda.c:36: check_error: Assertion `0' failed.
      

      • yolov3.cfg의 내용 변경

        • Before

        batch=1
        subdivisions=1
        width=416
        height=416
        

        • After

        batch=64
        subdivisions=16
        width=608
        height=608