Day & Time
18th September 2015, 15:00 ~
Room C816, Building of the Faculty of Science Graduate school of Science
Higher-level representation of local invariant features with applications in scene recognition, human action recognition and medicine
How to represent images and videos in a suitable way for recognition is one of the most important and fundamental research problems in computer vision. Recently significant progress has been made due to the development of local invariant features, which has led to many important applications, used by millions of people around the world every day. In this talk I will first provide a brief review of the basic ideas behind local feature design and implementation, and will then describe some recent work we have done to organize local features into higher-level representations, which leads to better recognition results. Some applications of the proposed approach to scene recognition, human actions recognition and in medicine will also be discussed.