The use of attention and spatial information for rapid facial recognition in video