Community based multi-group activity prediction and member identification

Background: Human activity recognition is a process by which a system can identify different human activities based on the observations on the individual's action and his/her surroundings. However, human beings are social by nature and like to build community depending upon a common point of interest and common characteristics. Therefore, for getting complete knowledge of an individual's activities, community based activity recognition - an extension of single activity recognition and group activity recognition is necessary. Additionally, for community activity monitoring, activity prediction is equi-important with activity recognition because early knowledge of the activities use to compare with the recognize activities for studying the certain change in the community activity pattern. We develop a recommendation system to identify the users based on their past and current performed activities. The system continuously identifies the users and generate alert to the user only if the system monitors any different activity pattern. The combination of the recommendation and alert system will as a whole monitor the individual's daily activity and track the irregularity in his/her life by comparing different activities over the time scale.