The test result is 0

[jack509513]

Thanks for your nice work!
We used the weights you provided for testing and the result was 0

[zhumanli]

Thanks for your nice work! We used the weights you provided for testing and the result was 0

Hi, I got the same 0 results. I'm wondering if you have resolved it. Thank you very much.

[Jeeseung-Park]

Sorry for the late reply.
I cannot reproduce this issue, can I get the exact environment and the script you runned?

[zhumanli]

Sorry for the late reply. I cannot reproduce this issue, can I get the exact environment and the script you runned?

I've used exactly the same scripts that you provided, but I didn't use Docker, instead, I followed the requirements of the nvcr.io/nvidia/pytorch:21.06-py3 Image file to create the env and installed other additional python libraries like gdown, einops, etc. In addition, since Step 3 (bash download/download_det_pose.sh) didn't work for me, I ran Step 4 and Step 5 for pose detection. Could it be because of the environment?

I got 0 results when I run the evaluation on the provided pre-trained weights, however, I got quite reasonable results for training a new network, even though the performance is a bit far from the reported ones. Thank you very much.

[Jeeseung-Park]

I wish I could see the results that pop up in the terminal after running the script. Thank you

[zhumanli]

I wish I could see the results that pop up in the terminal after running the script. Thank you
Hi,
Thanks a lot for your taking the time to look into this, here is the screenshot: