AI-Enabled Face Recognition based Contactless Attendance System
An easy to use, cost-effective, scalable contactless attendance system has been developed as alternate to Finger-print based attendance.
The face-recognition based solution comprises of Single Board Computer (SBC), display, webcam and a motion sensor. The application software front-end, as shown in figures has been developed using PyQt framework. to achieve the desired real-time performance.
Users have to wave their hand above a motion sensor (as shown in Fig 1(a)) to initiate the detection process. Once the person is identified, the attendance is marked without any other input. The face detection model is based on Multi-Task Convolutional Neural Network (MTCNN) and face recognition model uses contrastive learning method.
Screenshot of the Application Software and the deployed setup in CAIR
The system requires few (10-20) sample images of each employee to train the AI model. However, during the run-time, the system does not require any user face image. The training is done in an offline mode on a GPU compute platform. However, during operation the software runs on commodity CPU platform.
The system can be deployed to new places, which would require training of AI model using the face images of the employees at the place of deployment. New employees can also be enrolled easily and scaled up.