Human Gender Classification Based on Gait Features Using Kinect Sensor

Abstract

Gender classification is an important task in video surveillance. Terrorists and other criminals mostly act from a distance. Therefore, it is proposed that a gait-based method that recognizes humans remotely is required. In this paper, we propose a new method of gait-based gender classification based on the Kinect sensor, using a model based on the feature set, 'Dynamic Distance Feature (DDF)'. Nearest Neighbour (NN), Linear Discriminant Classifier (LDC), and Support Vector Machine (SVM) are used separately as a classification method. The method is tested based on skeletal data provided by the Microsoft Kinect. The experimental results show that the proposed method provided significant results by achieving 96.67%, 91% and 90% accuracy for gender classification using NN, LDC, and SVM respectively.