Getting wrong key-points detected from running test.py

Question

praptobala opened this issue 5 years ago · comments

Hi,
I ran "python -m torch.distributed.launch --nproc_per_node=1 test.py -i 1" successfully.

After finished running, I am reading 'results.json' file and plotting the keypoints with images using OpenCV.

But keypoints detected were not correct and its plotting multiple keypoints in one image.

Thanks,
Bala

fenglinglwb · Answer 1 · Wed Jul 31 2019 21:17:06 GMT+0800 (China Standard Time)

The input should be detected persons rather than original images. Is this right?

praptobala · Answer 2 · Thu Aug 01 2019 20:39:26 GMT+0800 (China Standard Time)

@fenglinglwb No, the input is original images. How can I input detected persons?

And I just want to detect one person from one image!

Linye Li · Answer 3 · Fri Aug 30 2019 11:45:46 GMT+0800 (China Standard Time)

you can crop person from image manually