CHEATING IN EXAM DETECTION SYSTEM USING GESTURE RECOGNITION BASED ON KINECT CAMERA

• Main Author: PUTRA (NIM: 13214031), HENDRA
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/27802
• Institution: Institut Teknologi Bandung

<p align="justify"> In facing the current era of globalization, we are required to be able to face increasingly competitive competition from within and also abroad. One way that we can compete is to improve the quality of human resources. Increased human resources can be done by improving the quality of existing education. The success of the quality of education can be measured through an evaluation conducted after a learning process conducted within a certain time frame. In general, evaluation is done through exams given at a certain period. But if seen, in fact the evaluation process by using the test or test still has many loopholes. The main gap is the oversight of the exam supervisor who is not careful and unfocused when supervising the exam. In addition to the use of time when the supervisor is being careless, cheating can also occur because the supervisor is less observant of the movements of the participants. Supervision by using technology has also begun to be applied, one of them by using surveillance cameras. Just like manual supervision, for long exam times the supervisors are also often bored so that there are times when the supervisor does not supervise the test participants so there is the possibility of cheating at that time. These gaps make the evaluation process no longer representative. To solve the problem was made a system that can detect the occurrence of fraud when the evaluation or exam process takes place. Detection system created using Kinect camera. Kinect cameras will capture the student skeleton and perform calculations against detected joints and can predict the movements performed by examinees. If a participant makes a suspicious move then the system will automatically notify the watchdog and take the captured image as evidence. The system is able to detect cheat gestures with a high degree of accuracy in realtime and is able to take pictures of gestures performed when cheating as evidence of automatic or manual cheating activity. <p align="justify">