An Implementation of an Automated Class Attendance System Based on Facial Detection and Recognition

Class attendance is a mandatory requirement amongst the TVET institutions in Kenya. Each institution has its own way of marking trainee’s attendance and keeping a record of the same. Majority of the institutions mark the trainee’s attendance manually while a few have adopted automated techniques such as voice recognition, eye detection, Radio frequency identification (RFID) or biometric. The daily maintenance of class attendance records is not only time consuming but also a difficult task. Therefore maintaining attendance register daily is a difficult and time-consuming task. Trainees have a tendency to manipulate the manual register by signing for their absent colleagues. Face detection and recognition has increased in the domain of image processing in the last few years and researchers have been able to implement it in various fields of our daily life such as for security purposes. Facebook for instance has implemented facial recognition algorithms into their website and applications, meaning that they cannot only find faces in an image; but they can also identify whose face it is as well. Facial recognition is an application of computer vision in the real world. This project aims at using a class video footage to extract picture snapshots, then detect and recognize faces from the snapshots. The detected faces will then be matched against registered records of faces in the database and mark students as being present or absent. This system enables trainers in learning institutions to mark learner’s class attendance easily and overcome the challenges of marking attendance manually. This paper has demonstrated a smart and efficient method for taking class attendance. The primary identification for human beings is the face. Therefore, face recognition offers an accurate way of overcoming ambiguities such as false attendance, time consumption and high cost. In addition, face recognition is a biometric method that has the merits of both low intrusiveness and accuracy