Cross-Age Face Recognition using Deep Residual Networks
Sarthak Babbar, Navroz Dewan, Kartik Shangle, Sudhanshu Kulshrestha, Sanjeev Patel · 2019
In the area of Artificial Intelligence and Deep Learning, face recognition is one of the demanding areas. Though a majority of research is going on, there is still a scope of improvement in the existing techniques. Our faces change with time when age increases, while the images in our dataset remain the same. We intend to study the accuracy of Residual Network (ResNet) [1] for the purpose of cross-age face recognition. The performance is compared to cross-age reference coding (CARC), Amazon Web Services (AWS) Rekognition and other techniques on the various data set viz., cross-age celebrity dataset (CACD) and a verification subset CACD-VS. ResNet and AWS Rekognition achieved 98.40% and 99.45% accuracy, respectively on the CACD-VS dataset.