Improved multiview biometric object detection for anti spoofing frauds

Improved multiview biometric object detection for anti spoofing frauds

12 February 2024 | P. Asmitha, Ch Rupa, S. Nikitha, J. Hemalatha, Aditya Kumar Sahu
The paper "Improved multiview biometric object detection for anti-spoofing frauds" addresses the challenges of human authentication, particularly in detecting and recognizing biometrics to combat spoofing, ID theft, and masking. The authors propose an enhanced version of ArcFace, called Improved ArcFace (I-AF), which combines Convolutional Neural Network (CNN) for feature extraction and RetinaFace for detection. This combination aims to improve accuracy and robustness in multi-face scenarios. The system is evaluated using real-time student data in a classroom to track attendance, achieving a 97% accuracy rate, which surpasses other methods. The research highlights the importance of addressing biases and security issues in facial recognition technology (FRT) and contributes to the development of reliable and safe identity verification methods, especially in crowded places. Key contributions include the analysis of specific encoding schemes and the design of a combination of DLA and RetinaFace for high accuracy in detecting multiple faces.The paper "Improved multiview biometric object detection for anti-spoofing frauds" addresses the challenges of human authentication, particularly in detecting and recognizing biometrics to combat spoofing, ID theft, and masking. The authors propose an enhanced version of ArcFace, called Improved ArcFace (I-AF), which combines Convolutional Neural Network (CNN) for feature extraction and RetinaFace for detection. This combination aims to improve accuracy and robustness in multi-face scenarios. The system is evaluated using real-time student data in a classroom to track attendance, achieving a 97% accuracy rate, which surpasses other methods. The research highlights the importance of addressing biases and security issues in facial recognition technology (FRT) and contributes to the development of reliable and safe identity verification methods, especially in crowded places. Key contributions include the analysis of specific encoding schemes and the design of a combination of DLA and RetinaFace for high accuracy in detecting multiple faces.
Reach us at info@study.space