DEff-GAN: Diverse Attribute Transfer for Few-Shot Image Synthesis (2023-02-28T00:00:00.000000Z)

TL;DR

The single-image GAN method is extended to model multiple images for sample synthesis and the discriminator is modified with an auxiliary classifier branch, which helps to generate a wide variety of samples and to classify the input labels.

Abstract

Requirements of large amounts of data is a difficulty in training many GANs. Data efficient GANs involve fitting a generators continuous target distribution with a limited discrete set of data samples, which is a difficult task. Single image methods have focused on modeling the internal distribution of a single image and generating its samples. While single image methods can synthesize image samples with diversity, they do not model multiple images or capture the inherent relationship possible between two images. Given only a handful of images, we are interested in generating samples and exploiting the commonalities in the input images. In this work, we extend the single-image GAN method to model multiple images for sample synthesis. We modify the discriminator with an auxiliary classifier branch, which helps to generate a wide variety of samples and to classify the input labels. Our Data-Efficient GAN (DEff-GAN) generates excellent results when similarities and correspondences can be drawn between the input images or classes.

Authors

Rajiv Kumar

2 papers

G. Sivakumar

2 papers

DEff-GAN: Diverse Attribute Transfer for Few-Shot Image Synthesis

TL;DR

Abstract

Authors

References29 items

Attribute Group Editing for Reliable Few-shot Image Generation

The MIS Check-Dam Dataset for Object Detection and Instance Segmentation Tasks

LoFGAN: Fusing Local Representations for Few-shot Image Generation

SA-SinGAN: self-attention for single-image generation adversarial networks

One-Shot GAN: Learning to Generate Samples from Single Images and Videos

Few-shot Image Generation via Cross-domain Correspondence

Repurposing GANs for One-Shot Semantic Part Segmentation

Learning Unsupervised Cross-domain Image-to-Image Translation Using a Shared Discriminator

Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis

MorphGAN: One-Shot Face Synthesis GAN for Detecting Recognition Bias

DeltaGAN: Towards Diverse Few-shot Image Generation with Sample-Specific Delta

F2GAN: Fusing-and-Filling GAN for Few-shot Image Generation

Differentiable Augmentation for Data-Efficient GAN Training

Improved Techniques for Training Single-Image GANs

Matchinggan: Matching-Based Few-Shot Image Generation

DAWSON: A Domain Adaptive Few Shot Generation Framework

Analyzing and Improving the Image Quality of StyleGAN

InGAN: Capturing and Retargeting the “DNA” of a Natural Image

Large Scale Adversarial Representation Learning

SinGAN: Learning a Generative Model From a Single Natural Image

FIGR: Few-shot Image Generation with Reptile

Deep Learning for Single Image Super-Resolution: A Brief Review

Zero-Shot Super-Resolution Using Deep Internal Learning

Improved Training of Wasserstein GANs

Generating Images with Perceptual Similarity Metrics based on Deep Networks

Deep Learning Face Attributes in the Wild

Smoothing the Generative Latent Space with Mixup-based Distance Learning

rcGAN: Learning a Generative Model for Arbitrary Size Image Generation

Few-shot un-sueprvised image-to-image translation

Field of Study

Venue Information

Name

Type

URL

Alternate Names