Attention-Aware Generative Adversarial Networks (ATA-GANs)

TL;DR

This paper introduces ATA-GANs, a novel GAN training method that leverages attention maps from a teacher network to enhance image quality and enable weakly supervised object localization, particularly for biomedical images.

Contribution

The paper proposes an attention transfer mechanism for GANs that improves image realism and provides attention maps for weakly supervised localization, a novel approach in GAN training.

Findings

Enhanced image quality with attention transfer

Effective weakly object localization in biomedical images

Discriminator focuses on regions of interest during training

Abstract

In this work, we present a novel approach for training Generative Adversarial Networks (GANs). Using the attention maps produced by a Teacher- Network we are able to improve the quality of the generated images as well as perform weakly object localization on the generated images. To this end, we generate images of HEp-2 cells captured with Indirect Imunofluoresence (IIF) and study the ability of our network to perform a weakly localization of the cell. Firstly, we demonstrate that whilst GANs can learn the mapping between the input domain and the target distribution efficiently, the discriminator network is not able to detect the regions of interest. Secondly, we present a novel attention transfer mechanism which allows us to enforce the discriminator to put emphasis on the regions of interest via transfer learning. Thirdly, we show that this leads to more realistic images, as the discriminator learns to put emphasis on the area of interest. Fourthly, the proposed method allows one to generate both images as well as attention maps which can be useful for data annotation e.g in object detection.