A Real-Time Automated Face Recognition and Detection System for Competitive Examination
Rajalakshmi Gurusamy, B. Ben Sujitha · Auerbach Publications eBooks · 2024
Current assessment systems primarily use biometric systems for candidate identification, recognition, and classification, as well as document image analysis tools. Unlike proposed paradigms that focus on images or videos for analysis, fraud is usually detected by analyzing document images. However, current assessment systems lack the ability to identify and verify candidate identities, making them susceptible to fraud. This study creates a deep learning model to perform automated face detection and identification to detect potential fraudsters in an examination environment. The model combines both physical and behavioral characteristics of individuals, such as facial features, fingerprints, iris recognition, and document image analysis. The proposed framework reveals tempering signatures that are undetectable to the human eye and reveals which parts of the face are considered relevant when viewing class activation maps. The model was tested on its own datasets, and results showed a high accuracy of 98.65% in identifying the correct candidate, with low false-positive and false-negative rates. It is more precise and resilient at different video processing speeds than previous techniques and is capable of accurately detecting spoofers in real-time with a high accuracy rate.