Department of Computer Engineering
S E M I N A R
Automatic Detection of Salient Objects for a Video Database System
Recently, the increase in the amount of multimedia data has unleashed the development of storage techniques. Multimedia database is one of the most popular of these techniques because of its scalability and ability to be queried by the media features. One downside of these databases is the necessity for processing of the media for feature extraction prior to storage and querying. Ever growing pile of media makes it harder this processing to be completed manually. This is the case with our Video Database System, BilVideo, as well. Improvements on computer vision techniques for object detection and tracking have made automation of this tedious manual task possible. We propose an automatic tool for extracting object's spatio-temporal features from videos. This presentation covers our scalable architecture for video processing and our methods for cut detection, object detection and tracking. For cut detection we use color histograms. Based on detected shots, our system detects salient objects in the scene making use of color regions and camera focus estimation. Then, we track detected objects based on their location, shape and estimated speed.
DATE: November11, 2005, Friday@ 13:40
PLACE: EA 502