Development of a real-time human-activity spotting system

Computer vision has brought about efficient human machine interaction and its area of research has been expanding. Human activity recognition is important in its application in surveillance systems, being able to effectively detect abnormal human motion through advanced recognition system. The objective of human activity recognition is to be able to recognize human motions and behaviour pattern in real-time. The aim is to be able to identify complex human activity so that it is able to replace humans in controlling surveillances system. It is complex to implement a real-time activity recognition system, therefore this project implemented a non-real time system to recognize human activity. Under controlled environments such as having a static background and an indoor testing implementation, we developed a human activity recognition system through combinations of methods such as active contour segmentation, capturing of human motion in MHI and performing LBP operation to conduct recognition. Finally, the performance of the system is thoroughly analyzed through the conduction of recognition test and rejection test.Bachelor of Engineerin