ARGUS activity recognition and object tracking based on multiple models

Home Research team Summary Tasks Publications Contacts Photos Grants Reserved Area

Argus rising, courtesy of David Curtis (web) See GRANTS for internships! There are hundreds of cameras deployed in public places (e.g., streets, shopping malls, airports) continuously acquiring video information. However, only a small fraction of the cameras outputs (<5%) is observed by humans or processed by computers. Most of this information is stored or destroyed without being watched or processed. There is currently a strong effort to bridge this gap. The project ARGUS aims at developing methods for the analysis and recognition of human activities using models learned from the video data.