An Effective Face Detection and Recognition Model Based on Improved YOLO v3 and VGG 16 Networks
Ali Nashwan Saleh, Alsafo Alaa Fares, Ali Hiba Dhiya, Taha Mustafa Sabah · International Journal of Computational Methods and Experimental Measurements · 2024
Face detection and recognition (FRD) technology is a very useful tool that involves taking pictures of people's faces and assessing their biological characteristics to compare and match facial data recorded in databases.Owing to its numerous advantages, including noncontact functionality, time and attendance tracking, medical applications and enhanced security and surveillance, this technology is finding increased application in a variety of contexts.Considering that the face images captured by these devices are influenced by many factors, such as light, posture, and backdrop environment, the recognition rate of current face recognition models remains inadequate.This paper presents a model that combines the You Only Live Once (YOLO) v3 algorithm for face detection with VGG16 networks for efficient face recognition.The model is specifically made to handle scenarios in which people share facial traits and to recognize people in various settings with accuracy.This paper uses two different public datasets to train and test the proposed model, WIDER FACE dataset for YOLO v3 and the Labelled Faces in the Wild (LFW) dataset for the VGG 16 networks, the improved network model performed better in identification and is more robust.Furthermore, the YOLO v3 network scored a little lesser accuracy of 95.9% in face detection, while the VGG 16 network achieved an amazing 96.2% accuracy in face recognition.