Key points are not available for this paper at this time.
A real-time life logging system that provide monitoring, recording and recognition of daily human activities using video cameras offers life-care or health-care services at smart homes. Such a vision-based life logging system can provide continuous monitoring and recording of a resident's daily activities from which one can obtain behavior patterns of daily life events and improve the quality of life especially for the elderly. This paper presents a real-time life logging system via depth imaging-based human activity recognition. A depth imaging device is utilized to obtain depth silhouettes of human activities. Then from the silhouettes, human body points information gets extracted and used in activity recognition, producing life logs. The system is composed of two key processes; one is training of the life logging system, and the other is running the trained life-logging system to record life logs. In the training process, the system includes the data collection from a depth camera, extraction of body points features from each depth silhouette and finally training of the activity recognizer (i.e., Hidden Markov Models). Then, after training, one can run the trained system which recognizes learned activities and store life logs in real-time. The proposed approach is evaluated against the life logging system that uses the conventional principal components (PC) and Radon transform features of depth silhouettes and achieves superior recognition rate. Real-time experimental results show the feasibility and functionality of the implemented system which could be used to generate life logs of human activities at smart homes.
Jalal et al. (Fri,) studied this question.