Affective Computing 

It aims to understand our emotion from images and videos, and makes robots communicate to us naturally. We mainly focus on Facial Expression Recognition, Multi-modal Emotion Recognition, Facial Action Units Detection, Emotion Perception problems. 

Video Action Detection 

It aims to understand what/when/where you do in videos. We mainly focus on spatial-temporal detection, temporal localization, online action detection/prediction, and online action start detection. 

Low-level vision 

The goal is to make the world look clearer! The area includes image/video super resolution, denoising, restoration and enhancement.