how can I prepare the labels for multi_pose training?

xingyizhou / CenterNet

Object detection, 3D detection, and pose estimation using center point detection:

MIT License

7.27k stars 1.93k forks source link

Open zry12321 opened 4 years ago

zry12321 commented 4 years ago

In fact, I can't understand the meaning of the variables in "multi_pose_dataset".

xingyizhou commented 4 years ago

Can you be more specific? Which variable?

zry12321 commented 4 years ago

Can you be more specific? Which variable? the variable 'kps' and 'kps_mask'.

xingyizhou commented 4 years ago

kps is the human keypoint location offset from its center, and kps_mask is if each keypoint is annotated. See here.