What Is Motion for?

In conjunction with ECCV 2022

Tel-Aviv, Israel


Motion is an important cue for many perception tasks from navigation to scene understanding. This workshop will explores various ways of representing motion information and set some future directions for the use of motion in Computer Vision. Traditionally, motion mainly refers to optical flow which has assisted several perception tasks. Recently, deep learning methods have led to several breakthroughs in motion estimation either explicitly with architectures that are designed to predict motion or implicitly where the motion information is captured in the learned representations, e.g. with 3D convolutions. This workshop has invited the researchers who contributed to these breakthroughs to show the role of motion in different sub-fields and hear what role they think that motion has to play in the future of Computer Vision.