Exploring methods and systems for vision based human activity recognition

摘要

This paper provides a comprehensive survey on the recent techniques of human activity recognition. The goal of the activity recognition is to automatically analyze the ongoing events. The applications of activity recognition are manifold, ranging from visual surveillance to control and video retrieval. The task is challenging due to variations in recording settings of people, environment and scene. This paper covers all aspects of the general framework of human activity recognition and provides a detailed overview of benchmark databases and current advances in this field. Finally, future directions to work on for this application are suggested.