Event Location: NSH 1507
Bio: Joseph Lim is a PhD student in the Department of Electrical Engineering and Computer Science at Massachusetts Institute of Technology (MIT), where he is advised by Professor Antonio Torralba. His research interests are in computer vision and machine learning. He has been motivated by the goal of building computer systems that automatically understand visual scenes in 2D and 3D. Joseph graduated with BA in Computer Science from UC Berkeley, where he worked under Professor Jitendra Malik. He was awarded the National Science Foundation Graduate Research Fellowship in 2009. More information can be found at his website: http://people.csail.mit.edu/lim

Abstract: Estimating the fine poses of the objects is a necessary step for robots, or for any visual system, to understand and manipulate objects in a scene. My recent work is about estimating pose of objects in 2D images at a very fine level of detail. Given the large search space of 3D poses, we directly use a 3D CAD model rather than a set of 2D images. I will first present how to utilize a compact representation of 3D CAD models by pose verification. Also, I will present a new method on leveraging additional data of 2D images with pose annotations. Based on the pose annotations, the method learns the part importance shared in 3D. Finally, I will go over the remaining challenges and possible future directions in fine pose estimation.