code can only detect one person in one image? about pytorch-cpn

[GengDavid]

Following the paper, this code is based on the top-down approach (detect person first then estimate each person's keypoints). If you want to use this code for multi-person images, you need to provide person bounding box then combine the results together.

[aidarikako]

Following the paper, this code is based on the top-down approach (detect person first then estimate each person's keypoints). If you want to use this code for multi-person images, you need to provide person bounding box then combine the results together.

Thank you for your work and reply

[GengDavid]

Seems the problem has been solved. Close this issue.

[my-hello-world]

@GengDavid @aidarikako @mingloo
hello,why i got so large loss like:

    Total params: 104.55MB

Epoch: 1 | LR: 0.00050000
iteration 100 | loss: 362.8368835449219, global loss: 246.98593711853027, refine loss: 115.85093688964844, avg loss: 403.03418150042546

any advice?tks

[shelly-xue]

@my-hello-world hi i meet the same large loss with you. Can you tell me why this? thank you very much