Hierarchical recognition of activities of daily living using multi-scale, multi-perspective vision and RFID

Research on computer-based recognition of ordinary household activities of daily living (ADLs) has been spurred by the need for technology to support care of the elderly in the home environment. We address the issue of recognizing ADLs at multiple levels of detail by combining multi-view computer vision and radio-frequency identification (RFID)-based direct sensors. Multiple places in our smart home testbed are covered by distributed synchronized cameras with different imaging resolutions. Learning object appearance models without costly manual labeling is achieved by applying the RFID sensing. A hierarchical recognition scheme is proposed by building a dynamic Bayesian network (DBN) that encompasses both coarse-level and fine-level ADL recognition. Advantages of the proposed approach include robust segmentation of objects, view-independent tracking and representation of objects and persons in 3D space, efficient handling of occlusion, and the recognition of human activity at both a coarse and fine level of detail.