Tác giả: TS.Nguyễn Thị Thủy
Cơ quan: Dept. of Computer Science, Hanoi University
Địa điểm: Phòng họp A, tầng 2 nhà S4.
Thời gian: 10/11/2013 2:00:00 PM
Object recognition is the task of finding the kinds of objects in images or video. One of the critical step in object recognition is to discover robust features for image representation. The feature design is very challenging because images exhibit high variations, are highly structured, and lie in high dimensional spaces. Feature learning is attractive as it exploits the availability of data and avoids the need of feature engineering. Learning feature has become increasingly popular and effective for visual recognition. A variety of learning and coding techniques have been proposed and evaluated, such as deep belief nets, deep autoencoders, sparse coding, etc. Motivated by our project work and the current trend in the field, I will present a short talk on sparse coding, deep learning and some of the advanced approaches for visual object recognition.;