Video surveillance at Lee Wee Nam Library (LWNL)

Abstract

This project aims to develop a seat occupancy live monitoring system which helps students to find a study space more quickly and efficiently in Lee Wee Nam Library (LWNL). The seat availability is determined by human detection using existing Internet Protocol (IP) cameras/ Closed-circuit Television (CCTV) cameras in LWNL. A brief introduction on the project background, project objective and project scope are done in the first chapter. Literature review on technologies and algorithms used in this project, including Center and Scale Prediction algorithm and Convolutional Neural network are discussed in the second chapter. In the third chapter, methodology for the project is explained in detail. The proposed human detector is programmed in Python, while the interface to display the seat availability is in the form of web application coded in HTML, CSS, JavaScript and PHP. Conclusion and future work are included in the last chapter. Due to the model of cameras in LWNL, the coverage of the human detection is currently restricted to level 5 of the LWNL. The accuracy of seat availability is approximately 60%. Only human heads are detected while personal belongings found in the area of study space are not counted in detection result.