US Patent Application 18031564. INCREMENTAL 2D-TO-3D POSE LIFTING FOR FAST AND ACCURATE HUMAN POSE ESTIMATION simplified abstract

INCREMENTAL 2D-TO-3D POSE LIFTING FOR FAST AND ACCURATE HUMAN POSE ESTIMATION

Organization Name

Intel Corporation

Inventor(s)

Anbang Yao of Beijing (CN)

Yangyuxuan Kang of Beijing (CN)

Shandong Wang of Beijing (CN)

Ming Lu of Beijing (CN)

Yurong Chen of Beijing (CN)

Wenjian Shao of Shanghai (CN)

Yikai Wang of Beijing (CN)

Haojun Xu of Beijing (CN)

Chao Yu of Beijing (CN)

Chong Wong of Beijing (CN)

INCREMENTAL 2D-TO-3D POSE LIFTING FOR FAST AND ACCURATE HUMAN POSE ESTIMATION - A simplified explanation of the abstract

This abstract first appeared for US patent application 18031564 titled 'INCREMENTAL 2D-TO-3D POSE LIFTING FOR FAST AND ACCURATE HUMAN POSE ESTIMATION

Simplified Explanation

The patent application discusses techniques for estimating the 3D pose of an object from a 2D input image.

• The techniques involve adjusting an initial 3D pose by projecting it to a 2D pose projection and applying a residual regressor to features based on the 2D pose projection and the detected 2D pose.
• The techniques also involve combining the 3D pose increment from the residual regressor with the current 3D pose estimate.
• These techniques aim to improve the accuracy of 3D pose estimation from a 2D input image.

Original Abstract Submitted

Techniques related to 3D pose estimation from a 2D input image are discussed. Such techniques include incrementally adjusting an initial 3D pose generated by applying a lifting network to a detected 2D pose in the 2D input image by projecting each current 3D pose estimate to a 2D pose projection, applying a residual regressor to features based on the 2D pose projection and the detected 2D pose, and combining a 3D pose increment from the residual regressor to the current 3D pose estimate.