AUTOMATED ATTENDANCE SYSTEM USING FPGA BASED ON FACE RECOGNITION

Abstract

The conventional method of taking attendance is done manually by the teacher or the administrator which requires considerable amount of time and efforts also involving errors and proxy attendance. Automated systems involving use of biometrics like fingerprint, iris recognition, etc are well developed in the recent years. However, it is intrusive and cost required for deployment on large scale gets increased substantially. To overcome these issues, biometric feature like facial recognition can be used which involves the phases such as image acquisition, face detection, feature extraction, face recognition and eventually marking the attendance. In this project, we had tried to convey successful and efficient "Automated Attendance System Using FPGA Based on Facial Recognition". Here, we have used MTCNN for the detection of face in captured frame and FaceNet for features extraction. Then, cosine similarity method is used to check the similarity between embeddings i.e. features of captured faces and faces in database. If the similarity value is greater than the defined threshold, then the captured face is treated as known and information is saved into data storage (excel sheet or google sheet) with name, date and time. The goal of this design is to detect and recognize the face in a captured frame using MTCNN and FaceNet for attendance purposes using FPGA, where we have explored two approaches: PS+PS and PS+PL approach. We found that the accuracy of the PS-PL approach is lower than the PS-PS approach due to quantization, but there was an improvement in performance and a reduction in the latency of the attendance system.