Age classification using an optimized CNN architecture

Abstract

With growing data size in multimedia systems, the need for successful image classification and retrieval systems becomes vital. Nevertheless, the performance of such systems is still limited for real-world applications. In this paper, we propose an optimized Convolutional Neural Network (CNN) architecture for the age classification problem. In order to justify the structure and depth of the proposed CNN-based framework, comprehensive experiments on a number of different CNN architectures are conducted. Based on the fitness of the age classification results with respect to success-error ratios, training times, and standard deviations of success rates; using exact, top-3 and 1-off criterion, the CNN architecture involving 4 convolutional layers and 2 fully connected layers is found to be superior to the other CNN-based architectures with different number of layers. We evaluate our method on a face database consisting of more than 55,000 images. © 2017 Association for Computing Machinery.