Development of Detection System on Suspicious Behaviour during Exam



Development of Detection System on Suspicious Behaviour During Exam using Image Processing (D2S2BDE) is a project that detect student suspicious behaviour during exam. Students intend to cheat during exam and the invigilator did not notice the students who cheat. This project is meant to discourage student from cheating during exam. The objectives of this project are to detect student suspicious behaviour using images processing, to analyse the suspicious behaviour and save the images as evidence. The methodology use in this project is prototype model methodology. By using prototype model, error may be detected earlier before implementing it into the real environment. This project detects suspicious behaviour such as peeking, exchanging notes, and talking during exam by capturing the real-time video of them using the webcam camera. An analysis is to be conducted on the suspicious images data that have been trained. If the success rate is high, the particular suspicious behaviour will be captured as evidence. This project used Faster-RCNN and Yolov4 models, which are going to be compared among them in term of performance. These two models are trained in two different platforms; Faster-RCNN is trained locally in laptop while Yolov4 is trained in cloud platform. The development of the system is done using Acer Aspire 5 laptop. According to result of the study on related topic, it shows that cheating during exam among student is high and it becomes a concern towards student growth. Cheating habit is an unethical habit that should be fixed or prevented. Hopefully, by having this project running, the students can be more honest in answering the exam paper.