Comparative analysis of augmented datasets performances of age invariant face recognition models

The popularity of face recognition systems has increased due to their non-invasive method image acquisition, thus boasting the widespread applications. Face ageing is one major factor that influences performance algorithms. In this study, authors present a comparative study two most accepted and experimented datasets (FG-Net morph II). These were used simulate age invariant (AIFR) models. Four types noises added at preprocessing stage. addition noise stage served as data augmentation technique number sample images available for deep convolutional neural network (DCNN) experimentation, improved proposed AIFR model trait aging features extraction process. models are developed with pre-trained Inception-ResNet-v2 architecture. On testing comparing models, results revealed FG-Net more efficient over Morph an accuracy 0.15%, loss function 71%, mean square error (MSE) 39% absolute (MAE) -0.63%.