Age Progression and Regression with Spatial Attention Modules

Qi Li,Yunfan Liu,Zhenan Sun

DOI: https://doi.org/10.48550/arXiv.1903.02133

2019-10-07

Abstract:Age progression and regression refers to aesthetically render-ing a given face image to present effects of face aging and rejuvenation, respectively. Although numerous studies have been conducted in this topic, there are two major problems: 1) multiple models are usually trained to simulate different age mappings, and 2) the photo-realism of generated face images is heavily influenced by the variation of training images in terms of pose, illumination, and background. To address these issues, in this paper, we propose a framework based on conditional Generative Adversarial Networks (cGANs) to achieve age progression and regression simultaneously. Particularly, since face aging and rejuvenation are largely different in terms of image translation patterns, we model these two processes using two separate generators, each dedicated to one age changing process. In addition, we exploit spatial attention mechanisms to limit image modifications to regions closely related to age changes, so that images with high visual fidelity could be synthesized for in-the-wild cases. Experiments on multiple datasets demonstrate the ability of our model in synthesizing lifelike face images at desired ages with personalized features well preserved, and keeping age-irrelevant regions unchanged.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The problems that this paper attempts to solve are two main challenges in facial age - change simulation:

1. **Multi - model training**: In order to simulate different age mappings, it is usually necessary to train multiple models. This not only increases the complexity of training, but also leads to a waste of computing resources.
2. **Realism of generated images**: The realism of the generated facial images is affected by the variations of training images in terms of pose, illumination and background, etc., resulting in the generated images may not be realistic enough.

To solve these problems, the paper proposes a framework based on conditional generative adversarial networks (cGANs) to achieve both age - progression and age - regression simultaneously. Specifically, the main contributions of the paper include:

- **Unified framework**: A unified cGAN framework is proposed, using two sets of generators to handle age - progression and age - regression tasks respectively. The input is a facial image and a target - age condition, and the output is a realistic age - transformed facial image.
- **Spatial attention mechanism**: A spatial attention mechanism is introduced to limit image modifications to only the areas closely related to age - change, thereby reducing the possibility of introducing distortion and ghosting artifacts and improving the quality of the generated images.
- **Experimental verification**: Extensive experiments were carried out on three age databases, and the results show that this method can accurately synthesize facial images of the target age, and well preserve personal characteristics and keep age - independent areas unchanged.

Through these innovations, the paper provides an efficient and high - quality solution for facial age - change simulation.