function get_landmark_and_bbox leads to the OpenCV error: (-215:Assertion failed) !ssize.empty() in function 'resize'

[XieJJ99]

We're using video as the input and then encounter the following error.

Traceback (most recent call last):
File "/root/miniconda3/envs/musetalk/lib/python3.10/runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_globals, None,
File "/root/miniconda3/envs/musetalk/lib/python3.10/runpy.py", line 86, in _run_code
exec(code, run_globals)
File "/root/musetalk/scripts/inference.py", line 162, in
main(args)
File "/root/miniconda3/envs/musetalk/lib/python3.10/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
return func(*args, **kwargs)
File "/root/musetalk/scripts/inference.py", line 91, in main
crop_frame = cv2.resize(crop_frame,(256,256),interpolation = cv2.INTER_LANCZOS4)
cv2.error: OpenCV(4.10.0) /io/opencv/modules/imgproc/src/resize.cpp:4152: error: (-215:Assertion failed) !ssize.empty() in function 'resize'

We added some debug information, and then found it relates to the get_landmark_and_bbox function, sometimes the y1 position in the coord_list is negative, here's the code in inference.py:

            coord_list, frame_list = get_landmark_and_bbox(input_img_list, bbox_shift)
#......
            if crop_frame.size == 0:
                print(f"Frame empty: {x1},{y1}->{x2},{y2}")
                continue

sample output in x1,y1->x2,y2 format:
Frame empty: 612,-1->848,343
Frame empty: 612,-1->848,343
Frame empty: 612,-10->851,340
Frame empty: 612,-8->848,332
Frame empty: 615,-18->848,326
Frame empty: 615,-15->848,323
Frame empty: 617,-21->848,323
Frame empty: 617,-15->845,323
Frame empty: 620,-8->845,326
Frame empty: 617,-2->845,332
Frame empty: 620,-2->848,332
Frame empty: 603,-1->828,309
Frame empty: 603,-1->828,309
Frame empty: 603,-1->828,315
Frame empty: 603,-1->828,315
Frame empty: 606,-1->828,315
Frame empty: 612,-5->834,307
Frame empty: 615,-2->834,304
Frame empty: 615,-5->837,301
Frame empty: 617,-6->837,298