Human Activity Recognition for Office Surveillance

Abstract

Human activity surveillance video systems are gaining popularity in the field of computer vision due to user demands for security as well as their growing importance in many applications such as elder care, home nursing, and unusual event alarming. Automatic activity recognition is the key to video surveillance. This paper presents a method for human activity recognition in office surveillance videos using machine learning models including convLSTM, GRCNN and LRCN with three main steps: pre-processing, feature extraction and activity classification. The main targeted activities are walking, sleeping on desk, handshaking, typing, opening or closing door. Experimental results demonstrate the effectiveness of the proposed LRCN approach in accurately recognizing human activities in office surveillance videos with acceptable training and testing accuracy.