ABSTRACT

One of the key requirements for a student to participate in a course assessment at the University is classroom attendance. A manual attendance record during lectures typically occupies both the lecturer's and the student's productive time in the class. The face is a physical expression of one's identity. As a result, we presented a facial recognition-based automated student attendance system. Face recognition systems are extremely helpful in real-world applications, particularly in security control systems. In this project, we present a technique for automating the attendance system using facial recognition. For every human individual, the face recognition technique has shown to be the primary and most significant biometric source of identity.  OpenCv and deep learning algorithms, as well as tracking attendance using this face recognition was employed in this work.

The following is how the approach is put into action. The picture data is obtained from a live video stream, and the identified face is subsequently transformed and cropped away from the main image using a Caffe-based DL face detector. This crop picture is used to produce 128-d face embeddings to quantify a face using the CNN deep learning model. The 128-d face embeddings are then used to train a support vector machine, which gives rise to our facial recognition engine. To record attendance for each student, the engine recognizes student face data. Any student's attendance data would be saved in the database for later processing. A RFID identification system is utilized by lecturers to start and terminate courses.

The model was planned and implemented as a web-based platform, with Python serving as the primary language for developing the facial recognition engine. The system is user-friendly and outperforms conventional systems in terms of reliability, efficiency, and accuracy by around 97 - 99 percent. It makes a big contribution to resolving the attendance recording problem in Nigerian universities.