Gender classification for real-time audience analysis system
Vladimir V. Khryashchev, Lev Shmaglit, Andrey Shemyakov, Anton Lebedev · 2014
The system allowing to extract all the possible information about depicted people from the input video stream is discussed. As reported previously, the proposed system consists of five consecutive stages: face detection, face tracking, gender recognition, age classification and statistics analysis. The crucial part of the system is gender classifier construction on the basis of machine learning methods. We propose a novel algorithm consisting of two stages: adaptive feature extraction and support vector machine classification. Both training technique of the proposed algorithm and experimental results acquired on a large image dataset are presented. More than 90% accuracy of viewer's gender recognition is achieved.