Design and Build Class Attendance System in Institut Teknologi Bandung using Face Recognition Technology

• Main Author: Juan, Winaldo
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42490
• Institution: Institut Teknologi Bandung

Attendance is an activity that always done in class and an important aspect on learning process in class. Attendance is a simple metric that shows student’s will to study. Many kinds of attendance implemented in schools and university, such as attendance using signing, student card, fingerprint, etc. System like those are difficult to control. Classes in Bandung Institute of Technology are using signing method for attendance. This method is very risky and prone to human error, especially on administration process and student’s manipulation. This makes data integrity is disturbed and many data do not represent the real life data. Survey is made and distributed to student in Bandung Institute of Technology to prove the problem exists. Survey showed that about 73% of 63 respondents experienced data integrity problem because of administration and about 62% from 63 respondents manipulated attendance data that leads to data integrity problem. The goal of this final project is to design an attendance system using face recognition technology. Expectation of this final project is to design a mature attendance system and feasible to replace current system that are using signing method. Methodology that is used in this final project is system engineering. This methodology includes needs analysis, concept exploration, concept definition, advanced development engineering design, dan integration & evaluation. This system has many components and this final project chose a right component by using analysis and clear arguments. Concept of this system is made clear on concept definition section. System consist of subsystems and subsystems consist of components. Concept definition discuss about chosen concept, such as camera placement, face recognition methods, and face gathering methods. This concept is the base of choosing components. Algorithm chosen is FaceNet for face recognition and support vector machine. Library chosen is Openface because it can accommodate system’s needs. Testing is done by following testing scenario and every scenario has its own goal to achieve. Testing is done in class so that it represent the real life class. The result of testing shows that this system is not perfect, but it can answer the main problem. Face recognized is 91 % of 50 faces that distributed on 10 pictures. Result of attendance simulation is system can identify every person and correct on identifying every person who attended class. Face recognition on 60 faces in one pictures only need 3 seconds. System succeed to detect all 60 faces in one frame. Designed system can get rid of human error in administration process and difficult to manipulate this system.