Face recognition through regional weight estimation

Face recognition has become a very important field of AI, with many competing techniques, both holistic and local. Recently, a new framework for embedding holistic face recognition algorithms into a regional voting approach, has been shown to be a very stable and accurate mechanism for face recognition. A new system is proposed, which extends the regional voting concept and adds weights to each region. Several techniques for estimating the weights are discussed. The system is shown to outperform several other leading face recognition algorithms.