Many research works on face recognition-based attendance systems have employed either a single camera or multi camera architecture in an attempt to capture and produce images that cover the entire participants of a class. Single stationary camera architectures have shown to have blind spots. This paper proposes a cost-effective single camera rail system that can be adaptive and implemented with any face recognition system designed to greatly improve audience coverage. It can be deployed in various audience locations and managed centrally. It also can be implemented in either wired or wireless modes. The system is a microcontroller-based (Embedded Systems Programming (ESP) microcontroller ESP 8266) system that consists of a single camera placed on a rail to allow for capturing of images of different sections of a class. The micro-controller determines each stop position of the camera by calculating the stop positions based on the configuration data (field of view angle, the maximum object distance and the rail length) stored in the MYSQL database. The stop positions are each passed to the stepper motor to pull the camera on the rails with the aid of a beaded string. The camera feed is captured programmatically using the Open-Source Computer Vision Library (Open CV). The frame at each stop position is automatically saved within a folder created for the class. Test results show that the entire participants of a class were covered and captured with stop positions calculated by the system and depending on the size of the class, the system can comfortably calculate stop positions that can cover the class attendees and the images captured was successfully transferred to the face recognition system for further processing.
Essien et al. (Thu,) studied this question.