最新消息

2016年05月20日学术报告:Understanding Tools: Task-Oriented Object Modeling, Learning and Recognition


添加时间:2016-05-18 16:05:50



 题目:Understanding Tools: Task-Oriented Object Modeling, Learning and Recognition

时间:2016年05月20日(周五),上午十点

报告人:Yixin Zhu

地点:研究生楼101

 

He is a 3rd year Ph.D student in Prof. Song-Chun Zhu‘s VCLA group of Statistics Department at UCLA. His research interests lie on computer vision & graphics, robotics, virtual reality and cognitive science, with the focuses on functional object and scene understanding. In particular, his ongoing work tries to integrate high-level cues (functionality, physics, etc) into computer vision to enable richer representations and deeper understandings.

 

He presents a new framework: task-oriented modeling, learning and recognition which aims at understanding the underlying functions, physics and causality in using objects as “tools”. Given a task, such as, cracking a nut or painting a wall, he represents each object, e.g. a hammer or brush, in a generative spatiotemporal representation consisting of four components: i) an affordance basis to be grasped by hand; ii) a functional basis to act on a target object (the nut), iii) the imagined actions with typical motion trajectories; and iv) the underlying physical concepts, e.g. force, pressure, etc. In a learning phase, our algorithm observes only one RGB-D video, in which a rational human picks up one object (i.e. tool) among a number of candidates to accomplish the task. From this example, the algorithm learns the essential physical concepts in the task (e.g. forces in cracking nuts). In an inference phase, the algorithm is given a new set of objects (daily objects or stones), and picks the best choice available together with the inferred affordance basis, functional basis, imagined human actions (sequence of poses), and the expected physical quantity that it will produce. From this new perspective, any objects can be viewed as a hammer or a shovel, and object recognition is not merely memorizing typical appearance examples for each category but reasoning the physical mechanisms in various tasks to achieve generalization.

 



作者:宋浩