Unattended Object Detection and Tracking

摘要

This paper proposes a design of smart surveillance system that can detect unattended objects and tracking owner of object upon the successful implementation, it can decrease the workload of security staff. Developing system is implemented by using image processing techniques alone with machine learning. In the event of an unattended object being detected, the system will alert the responsible person and display tracking the owner of the object. The system consists of four major steps: 1.) Image Acquisition 2.) Image Detection 3.) Event Recognition 4.) Result Visualization. The experiment is conducted in order to measure the accuracy of each process in the system. This system uses two datasets which is PETS 2006 and self generated datasets. Finally, the correctness of the background subtraction process is 89.55%, the correctness of the tracking process is 87.76%, and overall correctness of the system is 78.54%.