Exploring object segmentation methods in visual surveillance for human activity recognition

Abstract

Human activity recognition in video camera is the prime research topic in computer vision and machine learning since last many years. Visual cameras have been used in public and private place like railway station, shopping malls, airport, offices, schools and university, etc. to recognize threat in the scene. The automated visual surveillance system will help to catch suspect in the scene, person identification in distance and re-identification, traffic management, sports, human computer interface, etc. Generally speaking, the human activity recognition in visual surveillance divided into following stages: object (human or vehicle) segmentation, feature extraction, object classification, object tracking, and activity recognition. Hence, the robust and object segmentation method in video camera is a very important phase because the rest of methods are strongly rely on it. In this paper, we have studied the various methods and/or algorithms of object segmentation (human). We will also discuss the strength and weakness of algorithms, complexities in activity understanding and identify the possible future research challenges.