An automatic system for multiple human tracking and actions recognition in office environment

This paper presents an automatic system that detects humans, tracks to maintain knowledge of up to 2 human objects' whereabouts in an office environment and recognizes their simple actions (such as walking, sitting down, getting up, squatting down and standing up, in both lateral and frontal views) in color video for activities logging. The detection, tracking and human motion classifications are automatic and suitable for use in smart room/office environment.