Integrating YOLO and Custom CNN for Enhanced Visual Identification with the Face Recognition Library

Face detection and recognition play pivotal roles in various applications, from attendance management to secure urban living. This paper introduces an enhanced approach that integrates latest YOLOv8 (You Only Look Once), a customdesigned deep Convolutional Neural Network (CNN), and the face_recognition library to advance face detection and recognition in both biometric and practical domains. The proposed system leverages the efficiency of YOLOv8 for real-time multi-object detection, providing a robust foundation for identifying faces in diverse and dynamic environments. YOLOv8’s ability to process images at an impressive speed enhances the system’s responsiveness and adaptability, crucial for real-world applications. The integration of a custom-designed deep CNN, in conjunction with the face_recognition library, serves as the backbone for intricate feature extraction. This synergy enables high-precision face recognition even in challenging scenarios. The custom model’s adaptability to specific characteristics present in diverse face datasets, combined with the capabilities of the face_recognition library, enhances the system’s robustness and accuracy in recognizing faces with varying attributes. To evaluate the system’s performance, we conducted a various assessment using custom datasets representing real-world scenarios. The proposed system offers practicality in deployment. Its real-time capabilities make it suitable for time-sensitive applications, such as access control systems and security in urban environments. The integration of YOLOv8 with a custom deep CNN, and the face_recognition library represents a significant advancement in the field of face detection and recognition, offering a reliable and efficient solution for various challenges. As a comprehensive approach, this research contributes to the broader landscape of biometric technology, paving the way for enhanced face recognition systems applicable in various domains. The adaptability, accuracy, and efficiency demonstrated by our approach, utilizing both a custom-designed deep CNN and the face_recognition library, make it a promising candidate for integration into real-world applications, facilitating safer and more secure urban living environments and others.