Optimization of Resnet50 using Random Search for Face Recognition

Abstract

This study used the ResNet50 architecture and optimum configurations performed by Random Search to provide the values of an exhaustive investigation into the field of face recognition using Transfer Learning approach. The model's outstanding capability to succeed in an image classification task is demonstrated by its performance measures, which include precision (99.60%), recall (99.58%), and F1-score (99.59%).These findings highlight the importance of hyperparameter optimization as well as the capability of well-structured deep-learning models to produce astounding levels of accuracy and reliability