Real-time single-view video event recognition in controlled environments

This paper presents a real-time video event recognition system for controlled environments. It is able to recognize human activities and interactions with the objects of the environment by exploiting different cues like trajectory analysis, skin detection and people recognition of the foreground blobs of the scene. Time variations of these features are studied and combined using Bayesian inference to detect the events. Contextual information, including fixed objects' location, object types and event hierarchical definitions, is formally included in the system. A corpus of video sequences has been designed and recorded considering different complexity levels for object extraction. Experimental results show that our approach can recognize five kinds of events (two activities and three human-object interactions) with high precision operating at real-time.