Non-Intrusive Laboratory Attendance Confirmation via Object Detection using YOLO

The department of electrical engineering, State University of Malang has several laboratories equipped with CCTV as the monitoring tool. By adopting prototyping research method, the research is aimed to develop a device to detect the laboratory visitors’ attendance by implementing smart, nonintrusive, and automatic CCTV model. Different with RFID and biometric technology, this CCTV is designed without involving human in its operation. Identification of human presence, lecturers’ and students’, is carried out by computer vision techniques by using the YOLO library and rule-based method. With this non-intrusive detection, the machine will perform the attendance process that runs its operation continuously. This device ensures that lecturers and students are present throughout the designated laboratory schedule. Therefore, the information accumulated is valid for reporting the teaching and learning activity in the laboratory. As such information is also valuable for the quality assurance team in terms of analysis and improvements necessity for learning activities in the laboratory. The result of predictive model of human attendance in a laboratory reached the accuracy more than 95%. Meanwhile, the detection of lecturers and students has been successfully carried out based on the position of the human object on the x and y axes. Hence, this human object prediction model in laboratory X could be applied at other laboratories when these conditions are met: 1) The location of CCTV is the same as that of laboratory X’s; and 2) The room lighting setting is identical. Attendance monitoring continuously runs where information from CCTV is in form of static image for each specified period. Thus, attendance validity is the average detected presence of human objects for the duration of the activity in the laboratory.