Learning and Recognition for Assistive Computer Vision