LIFADER: Light Face Detection and Recognition approach for people tracking

Video surveillance is an inescapable tool in the recent modern world due to it importance in securing persons and goods. This gives the possibility to analyze information by processing multimedia signal and more precisely video signal. However, it come with its share of demands in the sense that processing of video signal uses more resources than normal information. In addition, the recent techniques used in detection and recognition tasks are machine learning/ deep learning approaches which is also demanding in terms of resources. On the other hand, computations in sensor network are subject to limitations in resources (memory in particular). This leads to challenges if an attempt is made to handle situations such as the monitoring of people in a quarantine zone for example. We propose in this paper an approach to solve the previously mentioned issues. The proposed approach combines an improved foreground segmentation approach a face detection and recognition approach as well a super resolution approach. This is with the intuition that in video surveillance, the processing of video information is necessary only when people move in the scene controlled by multimedia sensor network.