Swin-JDE icon indicating copy to clipboard operation
Swin-JDE copied to clipboard

Published 20 hours ago • verified publisher icon JackWoo0831

Datasets without id annotations?

Open yanghaibin-cool opened this issue 2 years ago • 0 comments

@JackWoo0831 Hi,Thank you for making your code open source！ I would like to ask how the nID is calculated for datasets without id labels, such as crowdhuman? How is it used to train reid? Where is it ignored when calculating id_loss? Could you point it out to me?Thank you for your reply.

yanghaibin-cool avatar Jun 12 '22 06:06 yanghaibin-cool