As a follow-up of the past IEEE-published workshops on Face Processing in Video: FPiV'04 (jointly with CVPR’04) and FPiV'05 (jointly with CRV’05), the First International Workshop on Video Processing for Security (VP4S-06) has been held jointly with the Canadian Conference on Computer & Robot Vision (CRV'06) on June 7-9 in Quebec City, Canada. This workshop kept its focus on processing video data (in particular such as coming from surveillance cameras),  while its main theme shifted towards security-related applications, and its interest extended from face detection, tracking, recognition, coding to people, objects, scene and action detection, tracking and recognition.

 

Attended by over a hundred researchers, VP4S-06 workshop consisted of two oral sessions and a poster/demo session.   Oral sessions featured seven papers accepted from fifteen Tier I submissions. These papers are published by IEEE as part of the CRV'06 proceedings. They include three papers on detecting and tracking people in a single camera video - one uses pixel-weighting to make object matching distance invariant, one applies K-means clustering to improve foreground detection, and one improves multiple-part object tracking by using correlograms; two papers on multiple-sensor setups for person – one improves person detection by using several cameras, the other presents a system that combines video data with the data obtained from infrared and fingerprint readers;  and two papers on  face recognition in video:  one tackles this most challenging video recognition problem by using stereo video, while the other does it by combining a global facial classifier with four local ones, corresponding to eyes, nose, and mouth.

 

The poster/demo session presented nine contributions accepted from the late Tier II submissions. These contributions are published at the workshop website and include posters and demos on  the following topics: multiple-object tracking with moving cameras using active contours; multiple-people tracking using correlation; comparison of mpeg-based motion detection with pixel-based motion detection;  designing  surveillance systems for long-term monitoring and activity summarization; using associative neural networks as alternative to histograms and correlograms in multiple-object tracking; video-based face characterization using factorized feature points; automated surveillance using self-organizing maps; fall detection using head trajectory analysis; and hand tracking based analysis of medication intake.

 

The success of the  workshop was possible due to the effort of the international program committee, additional reviewers, the organizers of the CRV’06 and its other joints conferences on Artificial Intelligence (AI’06) and Graphics Interface (GI’06) as well as the IEEE publisher. For more information on the VP4S-06  workshop, including the workshop online submissions and information on the next workshop venue and date, see the workshop website at www.computer-vision.org/4security.

.

Workshop Report:  VP4S-06