|
|
Automatic Image/Video Tagging
Dr. Shih-Fu Chang
Wednesday, February 21, 2007
2:00PM ~ 3:15PM Harris Center - 125
Abstract
Tagging image and video content with descriptive keywords has been the most intuitive way of accessing large collections of image/video data. The semantic tags, organized in various categories such as people, locations, events, and objects, can be used to answer user needs in searching, navigation, and editing. However, manual tagging is known to suffer from several problems – time consuming, error-prone, and lacking motivation, making it impractical in large-scale applications. In this talk, I will first briefly review efforts in improving image tagging by exploring web-based social networks or game-based systems. Then I will focus on recent development of large-scale visual concept ontology and various statistical methods for recognizing a large number (374) of semantic visual concepts. I will use DTO LSCOM visual ontology and NIST TRECVID evaluation as examples. The former involved definition of about 1000 visual concepts and annotations over 62,000 video shots. I will discuss the impact of using such automatic concept detectors on video search, and present the optimal strategies for fusing metadata from different modalities (text, visual, and concept). Besides the broadcast news domain used in TRECVID, I will present some preliminary results in developing a visual lexicon and automatic detectors in the consumer video domain.
Short Bio
Shih-Fu Chang is a Professor of Electrical Engineering and currently directs Digital Video and Multimedia Lab at Columbia University. He has made significant contributions in image/video search, media forensics, adaptive video transmission, and development of MPEG-7 international standards. His research has been among the most frequently cited in content-based image retrieval, with several large demo systems deployed such as VisualSEEk and VideoQ. He has been recognized with several awards, including a Navy ONR Young Investigator Award, an IBM Faculty Development Award, an ACM Recognition of Service Award, and an NSF CAREER Award. He and his students have received two Best Paper and six Best Student Paper Awards from IEEE, ACM, and SPIE. He served as a general co-chair of ACM Multimedia Conference 2000 and IEEE Multimedia Conference 2004. He has worked in different advising/consulting capacities for IBM, Microsoft, Kodak, PictureTel, and several other research institutions. He is an IEEE Fellow and currently Editor-in-Chief for IEEE Signal Processing Magazine.
http://www.ee.columbia.edu/dvmm
|
|
