Revolutionizing Real-Time Face Recognition: A Comparative Study of YOLOv5 against Traditional Methods
Romayyah Khan, Institute of Computer Sciences and Information Technology (ICS/IT), The University of Agriculture, Peshawar - Pakistan.
Imran Uddin, Institute of Computer Sciences and Information Technology (ICS/IT), The University of Agriculture, Peshawar - Pakistan.
Corresponding Author:
Romayyah Khan (romikhan944@gmail.com)
Abstract:
Face recognition systems requires high accuracy and efficiency in real time. The study presents a comparative analysis of yolov5, a state-of-the-art object detection model, against with the traditional face recognition models like CNN and FaceNet. While evaluating their performance based of accuracy and processing speed. All the models are trained and tested on a custom dataset, comprising images of 10 distinct individuals. For yolov5 the images were annotated with bounding boxes for faces. The results shows that yolov5 outperform against the traditional face recognition model in both accuracy and speed, achieved 99% accuracy for face recognition task with processing time as low as 2ms per step. The research provides important insights for improving face recognition technology and supports using YOLOv5 widely in real-world applications that need fast processing speed and accurate results.
Keywords:
Face Recognition; YOLOv5; CNN; FaceNet; Real-Time Processing.