Real-time face recognition in HD videos: Algorithms and framework

Computation of high dimensional data in real-time is a challenging task. In this paper, we consider the problem of detecting and recognizing faces in a real-time high definition video. We show time and accuracy of various incremental dimensionality reduction and incremental learning algorithms for face recognition and present a system architecture for implementing these algorithms. Further, we show that approximation in face detection can speed up the recognition time with minor loss in accuracy. We present our speedup results with the help of low cost GPUs. We also design a benchmark high definition multi subject video database and show our results over this benchmark.