Abstract

After decades of research and development, facial recognition system have attained considerable success in that many real-world applications are implemented, such as human computer interaction and multimedia communication. However, the technique of estimating human age via facial image is still a challenging problem for existing computer vision systems to automatically and effectively predict the exact age of humans. The problem derives from the fact that personalized aging patterns are temporal data and it is difficult to collect sufficient training data. In this paper, we utilize the classical Active Appearance Model and Artificial Neural Network to explore age range estimation using a Face a Person scheme in training phase. The experiments are conducted on the well-known aging database FG-NET. In order to show the effectiveness of the proposed method, the benchmark algorithm AGES, which needs subjects photo temporal sequences as the training set is used to compare against the proposed algorithm. The experimental results show that the performance of our method outperforms that of AGES algorithm.

Paper Details