Experience
Education
Bio
My research field is computer vision and robotics and in particular space and motion perception with machines. Since 1990, I have been studying multiple view geometry, image matching, stereo vision, recognition, sensor deployment, and camera design. Recently, my group has been working on: 3D Object and 3D Human Detection, Localization, and Reconstruction Starting with the assumption that we have only a database of 3D CAD models, we built a system for detection of 3D object categories in clutter and 3D localization of their instances (Zhu et al. ICRA 2014). We moved on to part-based representations detected by CNNs and proposed an algorithm for 3D reconstruction of object classes from single views (ICCV 2015), and for 3D human pose from monocular video (CVPR 2016). In (Zhou et al. CVPR 2015) we were able to show that 3D pose and shape can be obtained from any 2D projection using convex optimization. Earlier, in CVPR 2009, we had shown how to detect 3D objects in video using only CAD models and joint segmentation.