Look globally, age locally: Face aging with an attention mechanism

TL;DR

This paper introduces AcGANs, an attention-based GAN approach for face aging that improves image quality and identity preservation without pixel-wise loss, outperforming existing methods on the Morph dataset.

Contribution

The paper proposes an attention mechanism in GANs for face aging, reducing artifacts and preserving identity without pixel-wise loss, which is a novel approach.

Findings

Superior image quality and age accuracy on Morph dataset

Reduced ghost artifacts and blurriness in synthesized faces

Better identity preservation compared to existing methods

Abstract

Face aging is of great importance for cross-age recognition and entertainment-related applications. Recently, conditional generative adversarial networks (cGANs) have achieved impressive results for face aging. Existing cGANs-based methods usually require a pixel-wise loss to keep the identity and background consistent. However, minimizing the pixel-wise loss between the input and synthesized images likely resulting in a ghosted or blurry face. To address this deficiency, this paper introduces an Attention Conditional GANs (AcGANs) approach for face aging, which utilizes attention mechanism to only alert the regions relevant to face aging. In doing so, the synthesized face can well preserve the background information and personal identity without using the pixel-wise loss, and the ghost artifacts and blurriness can be significantly reduced. Based on the benchmarked dataset Morph, both qualitative and quantitative experiment results demonstrate superior performance over existing algorithms in terms of image quality, personal identity, and age accuracy.