Your Local GAN: Designing Two Dimensional Local Attention Mechanisms for Generative Models

by   Giannis Daras, et al.
The University of Texas at Austin

We introduce a new local sparse attention layer that preserves two-dimensional geometry and locality. We show that by just replacing the dense attention layer of SAGAN with our construction, we obtain very significant FID, Inception score and pure visual improvements. FID score is improved from 18.65 to 15.94 on ImageNet, keeping all other parameters the same. The sparse attention patterns that we propose for our new layer are designed using a novel information theoretic criterion that uses information flow graphs. We also present a novel way to invert Generative Adversarial Networks with attention. Our method extracts from the attention layer of the discriminator a saliency map, which we use to construct a new loss function for the inversion. This allows us to visualize the newly introduced attention heads and show that they indeed capture interesting aspects of two-dimensional geometry of real images.


page 2

page 5

page 8

page 9

page 12

page 16

page 17

page 18


Logo Generation Using Regional Features: A Faster R-CNN Approach to Generative Adversarial Networks

In this paper we introduce Local Logo Generative Adversarial Network (LL...

Self-Attention Generative Adversarial Networks

In this paper, we propose the Self-Attention Generative Adversarial Netw...

LOGAN: Latent Optimisation for Generative Adversarial Networks

Training generative adversarial networks requires balancing of delicate ...

Human Motion Modeling using DVGANs

We present a novel generative model for human motion modeling using Gene...

Attention-Aware Generative Adversarial Networks (ATA-GANs)

In this work, we present a novel approach for training Generative Advers...

Generating and Aligning from Data Geometries with Generative Adversarial Networks

Unsupervised domain mapping has attracted substantial attention in recen...

Gradient Layer: Enhancing the Convergence of Adversarial Training for Generative Models

We propose a new technique that boosts the convergence of training gener...

Code Repositories


[CVPR 2020] Official Implementation: "Your Local GAN: Designing Two Dimensional Local Attention Mechanisms for Generative Models".

view repo


Pytorch implementation of Your Local GAN (Two Dimensional Local Attention)

view repo


This repository provides GAN with various attention modules(Self-Attention, Criss-Cross-Attention, Your-Local-Attention).

view repo

Please sign up or login with your details

Forgot password? Click here to reset