Impersonation Attacks Detection in Online Exams Through Static Photo Analysis with Similarity Score

Academic integrity has become a major concern in online education, with impersonation attacks being a primary threat. Automatic proctoring has been developed to mitigate this issue, but security vulnerabilities related to impersonation detection using virtual webcams still exist. To address this issue and improve the reliability and accuracy of face verification technology, we propose a novel method that focuses on similarity scores generated from face recognition. Our proposed algorithm captures student activity during an exam every 5 seconds and compares it with the profile picture on the Learning Management System (LMS). The comparison results are converted into quantitative data and represented in a table and a graph. Each repetition of the similarity score values is compared with subsequent iterations over 15 iterations, automatically detecting potential impersonation cases and upholding the integrity of student assessments. The proposed algorithm ensures that a user is a valid user before activating the cheating detection pipeline, providing a comprehensive and efficient solution for detecting potential cheating incidents in online exams. This method represents a significant advancement in face recognition technology, with the capacity to confirm a user’s identity before activating the cheating detection pipeline. In conclusion, our proposed method provides a promising approach to combat impersonation attacks in online exams, ensuring academic integrity and fairness. By automating the process of comparing similarity score values, the proposed algorithm offers a rapid and reliable method for detecting potential impersonation cases and improving the effectiveness of automatic proctoring systems. This method can be integrated into existing systems to improve their effectiveness and ensure academic integrity in online education.