Multiple Incorrectly Detected Noses in 'result.png'

last-one / Pytorch_Realtime_Multi-Person_Pose_Estimation

Pytorch version of Realtime Multi-Person Pose Estimation project

MIT License

216 stars 70 forks source link

Multiple Incorrectly Detected Noses in 'result.png' #29

Open junzeliu opened 6 years ago

junzeliu commented 6 years ago

Hello,

I was running 'test_pose.py' with pytorch model, downloaded from GoogleDrive, based on 'ski.png'. However, there are hundreds of 'noses' detected on the image, while other body parts look fine. Either Python2.7 or Python3.6 ends up with this incorrect result. Could anyone give me any suggestions about this?

Thanks

puppywst commented 6 years ago

I guess you have taken the official openpose model. For official openpose, background is at the 18-th channel, nose at the 0-th channel, but in this repo, background is at the 0-th channel. Thus you get many noses (actually background points after nms).