In-Browser Attendance System using Face Recognition and Serverless Edge Computing

Abstract

Because of Covid-19, schools, colleges, and institutions have moved to online learning. The education system has encountered and continues to encounter various challenges in this online format in managing the attendance of the students. The teacher used to call out the students' roll numbers or names when they were in the physical education mode. Nowadays as the world is developing towards a digital era, numerous techniques of collecting attendance such as attendance via biometric technologies like eye recognition, face scanning, voice recognition, fingerprint analysis have earned a lot of fame. Face recognition is the most efficient of these approaches as the face can be captured using a camera and compared using a trained model, but the others are more complex to implement at the user end, and some even need hardware. A lot of research work has been already done related to face recognition using models such as YOLO, MTCNN, FaceNet, HOG, LBPH, C2D-CNN. Models are usually loaded in the backend which causes latency issues and makes the system inefficient to use. Our proposed system aims to perform face recognition within the browser itself with the help of serverless edge computing. For the students, a simple web portal is developed, from which they can navigate to our plugin extension, where the model will capture attendance and dynamically update it in a Google Sheet. Face detection was done with Tiny Face Detector, while face recognition was done with Face Recognition Net. A few more models operate in conjunction with these two, recognizing the student from his or her livestream, checking the student's authenticity using logged in credentials, and updating the attendance in real-time across the browser.