Assisted child-minding based on visual activity monitoring
Yong, Gary Wen Wei
Date of Issue2017-04-21
School of Computer Science and Engineering
Child-minding requires continuous high attention and awareness by the child minder that results in heavy fatigue after long periods of time. The objectives of this project are to implement a robust vision-based tracking capability and develop a semi-supervised approach specifying areas in the home that will trigger an alert to the adult to warn that the child has entered the danger zone, and to define or identify the child object model. The software to be developed is an interactive and semi-automated system based on surveillance systems which will assist the child minder in child-minding. The algorithms used are background subtraction which allows the object in motion to be detected and mean shift to track the object in motion using a tracker based on matching the most similar histogram values. The experiment conducted has successfully identified the object and can track the object in motion. The object is not 100% always in the bounding box, but after numerous tests, the tracker can track the object from the starting location to the its destination with prior precise definition of the object model. In short, it is able to track the child and alert the adult when the child enters the danger zone. This project has a very good market opportunity in the future as increasingly more parents are full-time working adults whom will require help in child-minding.
Final Year Project (FYP)
Nanyang Technological University