Age Invariant Face Recognition (AIFR)

Abstract

Facial recognition is employed in multiple methods in security situations such as surveillance, intelligence science, automatic image annotation, combating child trafficking and identifying criminals. Many of the facial recognition techniques developed in the past often show good results in confined space. However, many unexpected problems arise when applying these techniques to practical situations such as poor facial expressions, temporary disparities, and intentional or unintentional facial expressions in facial features. In addition, fluctuations in facial expression and the passage of time lead to the introduction of significant interclass variations, which also make facial recognition a challenging task. Data augmentation of age invariant face recognition (AIFR). Our procedure identify pictures through Viola Jones face detector and detects via the well-configured AIFR Convolutional Neural Network (CNN). During the transfer study, the pre-trained CNN learns the most consistent features of the years from the facial images of a few subjects to well-defined various ages’ facial expressions. We related the results of nine CNN 2Ds which were trained beforehand, at AIFR, having different sets of learning factors, based on the accuracy of sections and the timing of the selection of an accurate and fast and, effective model. Trials were made on the aging data-base of the AIFR face recognition and visual network (FG-NET-AD). The promising results obtained in this test indicate the effectiveness of our proposed model. In detailed comparisons made with modern AIFR methods, our proposed model excels in all aspects. The research work done in this thesis has also been developed to incorporate concealing flexibility and AIFR. For this purpose detailed tests have been performed on six challenging face shields and performance comparisons of four pre-trained CNNs to select the appropriate model for both AIFR and Disguise invariant face recognition (DIFR). Keywords: Convolutional Neural Network, Deep Learning, Flexible Face Recognition Age, Disguise Face Recognition, Enhanced Data.