报告嘉宾：加州大学圣地亚哥分校Hao Su博士（http://ai.ucsd.edu/~haosu/）
报告时间：2019年4月26日（周五）下午14:00-16:00
报告地点：北航新主楼G610
主办单位：北航计算机学院、虚拟现实技术与系统国家重点实验室
报告题目：Understanding the 3D Environments forInteractions
嘉宾简介：
HaoSu has been in UC San Diego as Assistant Professor of Computer Science andEngineering since July 2017. He is affiliated with the Contextual RoboticsInstitute and Center for Visual Computing. He served on the program committeeof multiple conferences and workshops on computer vision, computer graphics,and machine learning. He is the Area Chair of ICCV’19, CVPR’19, IPC of PacificGraphics'18, Program Chair of 3DV'17, Publication Chair of 3DV'16, and chair ofvarious workshops at CVPR, ECCV, and ICCV. He is also invited as keynotespeakers at workshops and tutorials in NIPS, CVPR, 3DV and CVPR, S3PM, etc. ProfessorSu is interested in fundamental problems in broad disciplines related toartificial intelligence, including machine learning, computer vision, computergraphics, robotics, and smart manufacturing. His work of ShapeNet, PointNetseries, and graph CNNs have significantly impacted the emergence and growth ofa new field, 3D deep learning. He used to work on ImageNet, a large-scale 2Dimage database, which is important for the recent breakthrough of computervision. Applications of Su's research include robotics, autonomous driving,virtual/augmented reality, smart manufacturing, etc.
报告摘要：
Beingable to understand the surrounding in both geometry and physics attributes aswe humans do is a key step for building intelligent autonomous agents. Thistalk will cover a series of research progress in my lab towards this direction,focusing on how machine learning, especially deep learning, can be used toaddress challenging problems in 3D reconstruction, semantic recognition, and mobilitystructure induction. In particular, I will focus on the understanding of objectparts. Object parts are handles of actionable information for interactionpurposes. Knowing such object part structure and being able to assembleactionable information on parts is thus fundamentally important. I will showhow this goal may be achieved by crowd-sourcing as well as algorithmicinduction efforts from daily observations. The content in the talk is basedupon latest papers published in SIGGRAPH Asia 2018 and CVPR 2019.
