Comparative Analysis of Efficient Face Recognition Models

This study presents a comparative analysis of four deep learning models, MobileNetV2, NasNetMobile,DenseNet169, and EfficientNetV2B0 applied to face recognition using transfer learning, L2 Regularization andDropout. The primary objective is to evaluate the models accuracy and training time. MobileNetV2 demonstrates asignificantly lower training time compared to other models while achieving 96% accuracy. EfficientNetV2B0, althoughattaining 99% accuracy, requires considerably longer training. In terms of overall efficiency, MobileNetV2 strikes thebest balance between accuracy and speed. Meanwhile, EfficientNetB0, though highly accurate, necessitateshyperparameter tuning to enhance its efficiency.CCS CONCEPTS • Computing methodologies ~ Machine learning ~ Machine learning algorithms ~ RegularizationAdditional Keywords and