deep_sort_pytorch
deep_sort_pytorch copied to clipboard
Published
20 hours ago •
ZQPei
ZQPei
About the indicator of reid model in evaluate.py
why dose the top1correct indicator can get from using res = scores.topk(5, dim=1)[1][:,0] as the code below？
res = scores.topk(5, dim=1)[1][:,0]
top1correct = gl[res].eq(ql).sum().item()
why not using res = scores.topk(1, dim=1)[1][:,0]
