Abstract:A Generative Adversarial Network (GAN) can learn the relationship between two image domains and achieve unpaired image-to-image translation. One of the breakthroughs was Cycle-consistent Generative Adversarial Networks (CycleGAN), which is a popular method to transfer the content representations from the source domain to the target domain. Existing studies have gradually improved the performance of CycleGAN models by modifying the network structure or loss function of CycleGAN. However, these methods tend to suffer from training instability and the generators lack the ability to acquire the most discriminating features between the source and target domains, thus making the generated images of low fidelity and few texture details. To overcome these issues, this paper proposes a new method that combines Evolutionary Algorithms (EAs) and Attention Mechanisms to train GANs. Specifically, from an initial CycleGAN, binary vectors indicating the activation of the weights of the generators are progressively improved upon by means of an EA. At the end of this process, the best-performing configurations of generators can be retained for image generation. In addition, to address the issues of low fidelity and lack of texture details on generated images, we make use of the channel attention mechanism. The latter component allows the candidate generators to learn important features of real images and thus generate images with higher quality. The experiments demonstrate qualitatively and quantitatively that the proposed method, namely, Attention evolutionary GAN (AevoGAN) alleviates the training instability problems of CycleGAN training. In the test results, the proposed method can generate higher quality images and obtain better results than the CycleGAN training methods present in the literature, in terms of Inception Score (IS), Fréchet Inception Distance (FID) and Kernel Inception Distance (KID).

CycleGAN with Better Cycles

ECycleGAN

An Extensive Study of Cycle-Consistent Generative Networks for Image-to-Image Translation

Unsupervised image-to-image translation via long-short cycle-consistent adversarial networks

Ecgan: Image Translation With Multi-Scale Relativistic Average Discriminator

Rethinking CycleGAN: Improving Quality of GANs for Unpaired Image-to-Image Translation

Cycle-Consistent Diverse Image Synthesis from Natural Language

ArCycleGAN: Improved CycleGAN for Style Transferring of Fruit Images

CycleSGAN: A cycle-consistent and semantics-preserving generative adversarial network for unpaired MR-to-CT image synthesis

CycleGAN with three different unpaired datasets

Optimization of image style transfer algorithm based on CycleGAN

Loss functions for Style Transfer with CycleGAN

UVCGAN: UNet Vision Transformer cycle-consistent GAN for unpaired image-to-image translation

Research on Image to Illustration Translation Method Based on CycleGAN

Cycle-Consistent Inverse GAN for Text-to-Image Synthesis

A Fully-Automatic Image Colorization Scheme Using Improved CycleGAN with Skip Connections

Asymmetric CycleGAN for image-to-image translations with uneven complexities

Image Translation with Attention Mechanism Based on Generative Adversarial Networks.

Realistic Image-to-Image Translation with Enhanced Texture

A Method based on Evolutionary Algorithms and Channel Attention Mechanism to Enhance Cycle Generative Adversarial Network Performance for Image Translation

Consistent Embedded GAN for Image-to-Image Translation.