An Industry-Ready Single PTZ-Camera Based Attendance Management System

摘要

In this paper, we present a smart management system for automatic headcount and attendance monitoring in the field of distance education and online training. Our system exploits the state-of-the-art object detection and face recognition models with customized modifications to achieve real-time performance under limited computation resources. Moreover, we present a novel route planning model for a pan–tilt–zoom camera (PTZ-camera) which significantly reduces the amount of time required to scan the whole environment. Experimental results demonstrate our system is able to deliver stable performance in terms of head detection and face recognition accuracy and system reliability. We believe our work showcases a successful integration of cutting edge academic approaches and affordable hardware resources, presenting an industry solution for the low-resolution face recognition task.