Human Identification in Video-streaming

Human identification is the trending feature in most software which is used by security organizations. Most of this type of software is oriented only for human identification by searching him from face database. We propose the software that identifies human in video-streaming by some classification parameters. In our work we classify people into age gender and race groups based on facial features. There are 3 face recognition algorithms: Eigenfaces Fisherfaces and Local Binary Patterns Histograms; that are suitable for classification task. For every algorithm classifiers with 1500 images from same dataset are trained. Purpose of the research is to determine the most appropriate one from chosen face recognition algorithms for human identification task based on classification parameters. For this purpose performance test experiment which determines the recognition time and rate of face recognition algorithms is performed. According to the results of experiment Fisherfaces is selected as most appropriate algorithm for our human identification task.