Fusion Network for Face-based Age Estimation

TL;DR

This paper introduces FusionNet, a CNN architecture that incorporates facial patches alongside the whole face to improve age estimation accuracy, demonstrating significant performance gains on the MORPH II dataset.

Contribution

The paper proposes a novel CNN architecture that emphasizes age-specific facial regions by integrating facial patches with the whole face for improved age estimation.

Findings

FusionNet outperforms state-of-the-art models on MORPH II.

Facial patches enhance age-specific feature extraction.

The approach significantly improves age estimation accuracy.

Abstract

Convolutional Neural Networks (CNN) have been applied to age-related research as the core framework. Although faces are composed of numerous facial attributes, most works with CNNs still consider a face as a typical object and do not pay enough attention to facial regions that carry age-specific feature for this particular task. In this paper, we propose a novel CNN architecture called Fusion Network (FusionNet) to tackle the age estimation problem. Apart from the whole face image, the FusionNet successively takes several age-specific facial patches as part of the input to emphasize the age-specific features. Through experiments, we show that the FusionNet significantly outperforms other state-of-the-art models on the MORPH II benchmark.