•  
  •  
 

Abstract

The Cognitively Motivated Video Detection (CMVD) system models our ability to detect interesting video events by first learning commonly occurring events in a particular environment. New events are then compared to previously learned events. If the event has not been seen before, the event is noted as interesting. Spatial- based events are described in terms of cognitively motivated low-level features including color, orientation, contrast, and motion. A schedule is used to represent temporal event patterns. The CMVD system has been successfully tested on video acquired from stationary cameras. The system accurately models human behavior.

Share

COinS