Abstract

3D HUMANS HUman pose Motion Activities aNd Shape in 3D

This workshop aims at gathering researchers who work on 3D understanding of humans from visual data, including topics such as 3D human pose estimation and tracking, 3D human shape estimation from RGB images or human activity recognition from 3D skeletal data. Current computer vision algorithms and deep learning-based methods can detect people in images and estimate their 2D pose with a remarkable accuracy. However, understanding humans and estimating their pose and shape in 3D is still an open problem. The ambiguities in lifting 2D pose to 3D, the lack of annotated data to train 3D pose regressors in the wild and the absence of a reliable evaluation dataset in real world situations make the problem very challenging. The workshop will include several high quality invited talks and a poster session with invited posters.