Hybrid Ontological Approach for Real Time Composite User Activity Recognition and Detection

Human activity recognition is an important parameter in the wide range of applications. To recognize the composite activities in the real and complex environments becomes difficult as monitoring the activities of multiple users cannot be done efficiently using conventional methods. The conventional methods mainly consist of knowledge driven and data driven approaches. In this paper a model for recognition of activities is proposed to detect the abnormal activities of the user. The model works in two modes i.e. learning mode and run mode. Initially, the model works on learning mode to get the sequence of activities and the detection of abnormal activities using hybrid ontological approach is done in the run mode. The simulation of this technique is presented and corresponding results are obtained to get the recognized activities. Thus, this paper calculates temporal relations between the series of actions termed into activity and later monitors this data with the real time data in the run mode.