Event Location: NSH 1507
Bio: Advised by Dr. Takeo Kanade, Jiyan Pan is currently a Ph.D. candidate in Robotics Institute at Carnegie Mellon University. His major research interest includes computer vision, machine learning, and artificial intelligence. His thesis research focuses on coherent scene understanding with 3D geometric reasoning. The goal is to develop a reasoning framework that, given an image, takes as input the noisy information from multiple sources and produces a geometrically coherent interpretation of the 3D world behind the image.

Abstract: In this talk, we present a coherent scene understanding algorithm that possesses those capabilities. In our approach, objects/surfaces are constrained by both global 3D geometries (such as gravity direction and ground plane), and local 3D geometries (such as depth ordering and space occupancy). We incorporate these two types of 3D geometric context in a RANSAC-CRF framework. More specifically, we use local objects/surfaces to propose hypotheses of global 3D geometries using a generalized RANSAC algorithm, and evaluate those hypotheses using a CRF which considers both the consistency of individual objects/surfaces under global 3D geometric context and the consistency between adjacent objects/surfaces under local 3D geometric context. We show that performing 3D geometric reasoning on both global and local levels greatly improves object detection and scene layout estimation.