Denoising 3D Human Poses from Low-Resolution Video using Variational Autoencoder
Chihiro Nakatsuka,Satoshi Komorita,Chihiro Nakatsuka,Satoshi Komorita
We tackle the problem of refining and denoising a series of 3D human poses estimated from a low-resolution video. Low-resolution often causes the wrong pose estimation, e.g., left-right switching and the absence of keypoints. We propose to use the variational autoencoder (VAE) to remove these challenging noises. The VAE model utilizes time-series information and motion priors in denoising. From ou...
