CMU-Perceptual-Computing-Lab / openpose

OpenPose: Real-time multi-person keypoint detection library for body, face, hands, and foot estimation
https://cmu-perceptual-computing-lab.github.io/openpose
Other
31.34k stars 7.87k forks source link

The poseKeyPoints is 3D or 2D data? #94

Closed segatecm closed 7 years ago

segatecm commented 7 years ago

I read the example code, tutorial_pose, and seen the code like this const auto poseKeyPoints = poseExtractorCaffe.getPoseKeyPoints();

so, please tell me the the poseKeyPoints is 3D or 2D data? if i want to get the 3D of human pose, how can i do it?

Thanks for your replay!

gineshidalgo99 commented 7 years ago

The output is 2D: (x,y,score).

There is no automatic 3D human pose so far: #38