An incremental learning method for face recognition under continuous video stream

摘要: The current technology in computer vision requires humans to collect images, store segment images for computers and train recognition systems using these images. It is unlikely that such a manual labor process can meet the demands of many challenging tasks. Our goal enable machines learn directly from sensory input streams while interacting with environment including human teachers. We propose new technique which incrementally derives discriminating features space. Virtual labels are formed by clustering output space extract organize resulting subspace coarse-to-fine fashion information decision tree. Such an incremental hierarchical regression (IHDR) tree be modeled probability distribution model. demonstrate performance algorithm on problem face video sequences 33889 frames length 143 different subjects. A correct rate 95.1% has been achieved.