Project introduction

Occluding contours are one of the important visual clues in computer vision. The local shape of an object’s occluding contour is known to constrain the local shape of the object’s surface qualitatively. Understanding the relationship between occluding contours and the 3D rim can assist us in multiple computer vision tasks, including depth estimation, among others. These projects have practical applications in real-life scenarios such as multi-view image synthesis or robotic grasping control.

3d rim vs 2d occluding contours

Our research aims to explore how common object regularities and projection rules might facilitate statistical shape estimations from occluding contours, even in the absence of strict quantitative links.