May I ask if you have encountered this error when building the network: a few targets cannot get positive sample points when they are processed by get_in_boxes_info, that is, all grids in this picture cannot get positive samples through calculation. If you encountered it, can you describe how you solved it?
yowo/matcher.py", line 191, in dynamic_kmatching
, pos_idx = torch.topk(
RuntimeError: selected index k out of range
May I ask if you have encountered this error when building the network: a few targets cannot get positive sample points when they are processed by get_in_boxes_info, that is, all grids in this picture cannot get positive samples through calculation. If you encountered it, can you describe how you solved it?
yowo/matcher.py", line 191, in dynamic_kmatching , pos_idx = torch.topk( RuntimeError: selected index k out of range
cost tensor([], size=(1, 0)) num_gtu: 1 gt_idx 0 cost[gt_idx] tensor([])