Human-Like Daily Action Recognition Model

This paper proposes a human-like action recognition model. When the model is implemented as a system, the system recognizes human actions similarly to human beings recognize. The recognition algorithm is constructed taking account of the following characteristics of human action recognition: simultaneous recognition, priority between actions, judgement fuzziness, multiple judge conditions for one action, and recognition ability from partial view of the body. The experiments based on a comparison with completed questionnaires demonstrated that the system recognizes human action the way like a human being does. Results ensure natural understanding of human action by a system, which leads to smooth communication between computer systems and human beings.