ENHANCING AGE ESTIMATION FROM SCLERA IMAGES USING RESNET-50, VGG16, AND RANDOM FOREST

Abstract

This study presents a novel hybrid model for age prediction from sclera images, combining deep learning architectures ResNet-50 and VGG-16 with a Random Forest classifier. The hybrid approach aims to optimize both accuracy and computational efficiency, addressing limitations in previous methodologies. Results demonstrate exceptional performance, with the hybrid model achieving an overall accuracy of 98.85% and outperforming benchmark models. Detailed evaluation metrics reveal high precision, recall, and F1-scores across age groups, supported by insights from the confusion matrix. The model's practical applicability is demonstrated through efficient training and testing processes. This research bridges gaps in existing literature by integrating transfer learning, deep learning, and ensemble methods, while also addressing issues of computational complexity. The study underscores the potential of hybrid models to advance age prediction from biometric images, setting a new benchmark for future research in the field.