RGBD object pose recognition using local-global multi-kernel regression

Abstract

The advent of inexpensive depth augmented color (RGBD) sensors has brought about a large advance-ment in the perceptual capability of vision systems and mobile robots. Challenging vision problems like object category, instance and pose recognition have all ben-efited from this recent technological advancement. In this paper we address the challenging problem of pose recognition using simultaneous color and depth infor-mation. For this purpose, we extend a state-of-the-art regression framework by using a multi-kernel approach to incorporate depth information to perform more effec-tive pose recognition on table-top objects. We do exten-sive experiments on a large publicly available dataset to validate our approach. We show significant perfor-mance improvements (more than 20%) over published results. 1.