Video-Base People Counting and Gender Recognition

摘要

Video based people counting and gender recognition are important but challenging tasks. A neural network method for video-base people counting and gender recognition is proposed in this paper. A multilayer perceptron structure is constructed and meaningful features from target video are extracted as input. The neural network is trained by back-propagation training algorithm. This method is experimented on four videos, including more than 240 peoples. Experiment results have shown the effectiveness of this method.