Multi-agent System for Tracking and Classification of Moving Objects

Abstract

Abstract. In the past, computational barriers have limited the complexity of video and image processing applications but recently, faster computers have enabled researchers to consider more complex algorithms which can deal successfully with vehicle and pedestrian detection technologies. However, much of the work only pays attention to the accuracy of the final results provided by the systems, leaving aside the computational efficiency. Therefore, this paper describes a system using a paradigm of multi-agent system capable of regulating itself dynamically taking into account certain parameters pertaining to detection, tracking and classification, to reduce the computational burden as low as possible at all times without this in any way compromise the reliability of the result.