In this paper, a child monitoring system is proposed to watch and track children while playing. In particular, the system detects the moving child using a background subtraction approach which relies on adaptive nonparametric Gaussian mixture modelling. Then, Kalman Filter is adopted for the tracking task. If the child oversteps the specified area or does something he is not supposed to do, the proposed system notifies the parents and convey a message to the child. The experimental results proved the effectiveness of the proposed system with an average F-score of 0.87 with a standard deviation of 0.07.
|