Modeling of cheating detection in distance education exams with gaze detection algorithms

Even though distance education gives students the opportunity to study in their own home environment; ensuring the security of distance exams is an ongoing problem of distance education. The development of automatic exam tracking and control models in distance education systems is of utmost importance for the reliable execution of distance education systems and reducing costs. In this study, it is aimed to create a cheat detection system model and to develop new methods, algorithms and models in order to provide successful results by using webcams connected to students' computers in distance education exams. In this thesis, two new methods and algorithms based on Viola-Jones and Facial-Landmarks in the field of eye detection from camera were developed and the system was optimized. The limitations of web cameras and their problems in finding views have been revealed and methods that can be used for solution have been determined. In order to determine the pupil positions, new methods based on template matching created specifically for the eye and unprecedented in the literature have been developed. Within the scope of the study, tests were made in three different databases and the results were examined. The new algorithm and model developed were compared with each other theoretically and practically and it was determined that the performance rate of the new model was high.