Accurate and efficient attendance management remains a significant challenge in educational institutions due to issues such as proxy attendance, manual errors, and time inefficiency. This paper proposes a multi-layer biometric attendance system based on computer vision techniques to address these limitations. The proposed system integrates face recognition as the primary identification method, enhanced with facial landmark validation, iris region localization, and basic anti-spoofing mechanisms to improve robustness, security, and accuracy.

The system is implemented using Python-based computer vision frameworks and operates on real-time video input for automated attendance marking. Facial embeddings are generated and matched against a structured database, while landmark-based geometric validation and iris region analysis act as additional verification layers. Furthermore, anti-spoofing measures such as live face verification (e.g., blink detection or facial movement analysis) are incorporated to prevent unauthorized access using photos or videos.