A Perspective way of designing Intelligent systems with Face Detection and Recognition using Artificial Intelligence for Authentication

Abstract

This project proposes an AI-based approach for face detection and recognition using YOLOv4 and CNN. It aims to be used in various applications like security systems and automatic attendance marking in educational institutions. Traditional machine learning methods like FaceNet and MTCNN have not achieved satisfactory accuracy. Hence, deep learning techniques are employed. YOLO, a deep learning library, is utilized for implementation. The key steps include creating an effective dataset, labeling images with faces using LabelImg software, and training the dataset using Google Colab GPU. The trained dataset is used to mark bounding boxes around faces in images or real-time video. The YOLOv4 technique is implemented using the darknet framework. Around 20 images are used for testing, resulting in an approximate accuracy of 85% due to the high accuracy of YOLOv4.