Learning to Discover Cross-Domain Relations with Generative Adversarial Networks

Abstract

While humans easily recognize relations between data from different domains without any supervision, learning to automatically discover them is in general very challenging and needs many ground-truth pairs that illustrate the relations. To avoid costly pairing, we address the task of discovering cross-domain relations given unpaired data. We propose a method based on generative adversarial networks that learns to discover relations between different domains (DiscoGAN). Using the discovered relations, our proposed network successfully transfers style from one domain to another while preserving key attributes such as orientation and face identity. Source code for official implementation is publicly available https://github.com/SKTBrain/DiscoGAN

Accepted to International Conference on Machine Learning (ICML) 2017

References in corpus (5)

• Conditional Generative Adversarial Nets
• Deep Generative Image Models using a Laplacian Pyramid of Adversarial Networks
• Unsupervised Cross-Domain Image Generation
• Disentangling factors of variation in deep representations using adversarial training
• Mode Regularized Generative Adversarial Networks

Cited by in corpus (190)

• Deep Face Recognition: A Survey
• COVID-19 Image Data Collection: Prospective Predictions Are the Future
• Model Adaptation: Unsupervised Domain Adaptation without Source Data
• U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation
• Generative Adversarial Networks in Computer Vision: A Survey and Taxonomy
• Federated Adversarial Domain Adaptation
• Generative Adversarial Networks (GANs) in Networking: A Comprehensive Survey & Evaluation
• Contrastive Learning for Unpaired Image-to-Image Translation
• A survey of synthetic data augmentation methods in computer vision
• StarGAN v2: Diverse Image Synthesis for Multiple Domains
• Cross-regional oil palm tree counting and detection via multi-level attention domain adaptation network
• Few-Shot Unsupervised Image-to-Image Translation
• Privacy Intelligence: A Survey on Image Privacy in Online Social Networks
• Detecting GAN generated Fake Images using Co-occurrence Matrices
• Translating and Segmenting Multimodal Medical Volumes with Cycle- and Shape-Consistency Generative Adversarial Network
• Facial-Sketch Synthesis: A New Challenge
• Moment Matching for Multi-Source Domain Adaptation
• Deepfake Detection Scheme Based on Vision Transformer and Distillation
• Image Captioning with Very Scarce Supervised Data: Adversarial Semi-Supervised Learning Approach
• Dual Adversarial Semantics-Consistent Network for Generalized Zero-Shot Learning
• Generative Adversarial Networks for Image and Video Synthesis: Algorithms and Applications
• Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection
• Physics-Based Generative Adversarial Models for Image Restoration and Beyond
• Low-light Image Enhancement Algorithm Based on Retinex and Generative Adversarial Network
• Self-Ensembling with GAN-based Data Augmentation for Domain Adaptation in Semantic Segmentation
• Mask-aware Photorealistic Face Attribute Manipulation
• An Efficient Solution for Breast Tumor Segmentation and Classification in Ultrasound Images Using Deep Adversarial Learning
• SPIGAN: Privileged Adversarial Learning from Simulation
• Learning Generative Models across Incomparable Spaces
• Data-Driven Approach to Encoding and Decoding 3-D Crystal Structures
• Multi-Channel Attention Selection GAN with Cascaded Semantic Guidance for Cross-View Image Translation
• Unpaired Speech Enhancement by Acoustic and Adversarial Supervision for Speech Recognition
• Cross-domain Correspondence Learning for Exemplar-based Image Translation
• Cali-Sketch: Stroke Calibration and Completion for High-Quality Face Image Generation from Human-Like Sketches
• Attribute-Guided Face Generation Using Conditional CycleGAN
• Similarity-preserving Image-image Domain Adaptation for Person Re-identification
• CrystalGAN: Learning to Discover Crystallographic Structures with Generative Adversarial Networks
• Structural-analogy from a Single Image Pair
• Fake Generated Painting Detection via Frequency Analysis
• Learning Layout and Style Reconfigurable GANs for Controllable Image Synthesis
• Image-to-Image Translation: Methods and Applications
• Towards Shape Biased Unsupervised Representation Learning for Domain Generalization
• GMM-UNIT: Unsupervised Multi-Domain and Multi-Modal Image-to-Image Translation via Attribute Gaussian Mixture Modeling
• Keep it Simple: Image Statistics Matching for Domain Adaptation
• CorGAN: Correlation-Capturing Convolutional Generative Adversarial Networks for Generating Synthetic Healthcare Records
• Attention-Guided Generative Adversarial Networks for Unsupervised Image-to-Image Translation
• Gesture-to-Gesture Translation in the Wild via Category-Independent Conditional Maps
• Dual Contrastive Learning for Unsupervised Image-to-Image Translation
• Generative Adversarial Networks (GANs): What it can generate and What it cannot?
• A Flexible Selection Scheme for Minimum-Effort Transfer Learning
• Medical Image Generation using Generative Adversarial Networks
• Reusing Discriminators for Encoding: Towards Unsupervised Image-to-Image Translation
• Anisotropic Stroke Control for Multiple Artists Style Transfer
• Style Example-Guided Text Generation using Generative Adversarial Transformers
• Training of deep cross-modality conversion models with a small dataset, and their application in megavoltage CT to kilovoltage CT conversion
• Image Morphing with Perceptual Constraints and STN Alignment
• Network Architecture Search for Domain Adaptation
• Attribute Guided Unpaired Image-to-Image Translation with Semi-supervised Learning
• Contextual colorization and denoising for low-light ultra high resolution sequences
• Cross-Domain Cascaded Deep Feature Translation
• Generative Feature Replay For Class-Incremental Learning
• MixerGAN: An MLP-Based Architecture for Unpaired Image-to-Image Translation
• The Spatially-Correlative Loss for Various Image Translation Tasks
• DAL: Dual Adversarial Learning for Dialogue Generation
• Adaptive Hierarchical Dual Consistency for Semi-Supervised Left Atrium Segmentation on Cross-Domain Data
• Learning Fixed Points in Generative Adversarial Networks: From Image-to-Image Translation to Disease Detection and Localization
• Detection, Attribution and Localization of GAN Generated Images
• Cascade EF-GAN: Progressive Facial Expression Editing with Local Focuses
• DeepKeyGen: A Deep Learning-based Stream Cipher Generator for Medical Image Encryption and Decryption
• Iconify: Converting Photographs into Icons
• Unpaired Image Enhancement Featuring Reinforcement-Learning-Controlled Image Editing Software
• Neural Rerendering in the Wild
• Multi-Mapping Image-to-Image Translation with Central Biasing Normalization
• Semi-supervised Feature-Level Attribute Manipulation for Fashion Image Retrieval
• DG-Font: Deformable Generative Networks for Unsupervised Font Generation
• VoiceGrad: Non-Parallel Any-to-Many Voice Conversion with Annealed Langevin Dynamics
• SingleGAN: Image-to-Image Translation by a Single-Generator Network using Multiple Generative Adversarial Learning
• Recursive Chaining of Reversible Image-to-image Translators For Face Aging
• PlayVirtual: Augmenting Cycle-Consistent Virtual Trajectories for Reinforcement Learning
• Learning Cross-domain Generalizable Features by Representation Disentanglement
• DCDLearn: Multi-order Deep Cross-distance Learning for Vehicle Re-Identification
• COCO-FUNIT: Few-Shot Unsupervised Image Translation with a Content Conditioned Style Encoder
• GANHopper: Multi-Hop GAN for Unsupervised Image-to-Image Translation
• DUAL-GLOW: Conditional Flow-Based Generative Model for Modality Transfer
• Deep Adversarial Transition Learning using Cross-Grafted Generative Stacks
• Class-imbalanced Domain Adaptation: An Empirical Odyssey
• Handloom Design Generation Using Generative Networks
• Label-Noise Robust Generative Adversarial Networks
• Open-Edit: Open-Domain Image Manipulation with Open-Vocabulary Instructions
• Image Synthesis From Reconfigurable Layout and Style
• TSIT: A Simple and Versatile Framework for Image-to-Image Translation
• Unsupervised Domain Adaptation for Visual Navigation
• CycleGAN-VC2: Improved CycleGAN-based Non-parallel Voice Conversion
• Multi-Target Domain Adaptation via Unsupervised Domain Classification for Weather Invariant Object Detection
• Exploring Explicit Domain Supervision for Latent Space Disentanglement in Unpaired Image-to-Image Translation
• On Deep Domain Adaptation: Some Theoretical Understandings
• Potential Flow Generator with $L_2$ Optimal Transport Regularity for Generative Models
• Unsupervised Enhancement of Real-World Depth Images Using Tri-Cycle GAN
• Diverse Semantic Image Synthesis via Probability Distribution Modeling
• toon2real: Translating Cartoon Images to Realistic Images
• Asymmetric GANs for Image-to-Image Translation
• Learning to Transfer: Unsupervised Meta Domain Translation
• GA-DAN: Geometry-Aware Domain Adaptation Network for Scene Text Detection and Recognition
• TuiGAN: Learning Versatile Image-to-Image Translation with Two Unpaired Images
• Multi-Domain Image Completion for Random Missing Input Data
• Dual Adversarial Inference for Text-to-Image Synthesis
• Generative Adversarial Network with Multi-Branch Discriminator for Cross-Species Image-to-Image Translation
• Dual Generator Generative Adversarial Networks for Multi-Domain Image-to-Image Translation
• Disentangled Cycle Consistency for Highly-realistic Virtual Try-On
• Adversarial-based neural networks for affect estimations in the wild
• Cycle Consistent Probability Divergences Across Different Spaces
• Unpaired Image-to-Image Translation using Adversarial Consistency Loss
• Point Cloud Completion by Learning Shape Priors
• Controlling biases and diversity in diverse image-to-image translation
• Crossmodal Voice Conversion
• Ancient Painting to Natural Image: A New Solution for Painting Processing
• Emotion Generation and Recognition: A StarGAN Approach
• Neural Topic Modeling with Cycle-Consistent Adversarial Training
• Attribute-guided image generation from layout
• Ear2Face: Deep Biometric Modality Mapping
• Learning a Domain-Agnostic Visual Representation for Autonomous Driving via Contrastive Loss
• GANs N' Roses: Stable, Controllable, Diverse Image to Image Translation (works for videos too!)
• Global and Local Alignment Networks for Unpaired Image-to-Image Translation
• Augmented Cyclic Consistency Regularization for Unpaired Image-to-Image Translation
• Improved BiGAN training with marginal likelihood equalization
• Integrating Categorical Semantics into Unsupervised Domain Translation
• Independent Encoder for Deep Hierarchical Unsupervised Image-to-Image Translation
• Spatial-Aware GAN for Unsupervised Person Re-identification
• Visual Diver Face Recognition for Underwater Human-Robot Interaction
• One-to-one Mapping for Unpaired Image-to-image Translation
• Mapping Low-Resolution Images To Multiple High-Resolution Images Using Non-Adversarial Mapping
• Symbiotic Adversarial Learning for Attribute-based Person Search
• Mix and match networks: cross-modal alignment for zero-pair image-to-image translation
• MU-GAN: Facial Attribute Editing based on Multi-attention Mechanism
• DLOW: Domain Flow for Adaptation and Generalization
• Bayesian Cycle-Consistent Generative Adversarial Networks via Marginalizing Latent Sampling
• Unsupervised Multi-Domain Multimodal Image-to-Image Translation with Explicit Domain-Constrained Disentanglement
• Show, Attend and Translate: Unpaired Multi-Domain Image-to-Image Translation with Visual Attention
• Risk Bounds for Unsupervised Cross-Domain Mapping with IPMs
• Mimicking the In-Camera Color Pipeline for Camera-Aware Object Compositing
• Nonparallel Voice Conversion with Augmented Classifier Star Generative Adversarial Networks
• MangaGAN: Unpaired Photo-to-Manga Translation Based on The Methodology of Manga Drawing
• Imaging radar and LiDAR image translation for 3-DOF extrinsic calibration
• LEED: Label-Free Expression Editing via Disentanglement
• MI^2GAN: Generative Adversarial Network for Medical Image Domain Adaptation using Mutual Information Constraint
• Cross Domain Knowledge Transfer for Unsupervised Vehicle Re-identification
• Duality Regularization for Unsupervised Bilingual Lexicon Induction
• One-Shot Image-to-Image Translation via Part-Global Learning with a Multi-adversarial Framework
• Augmenting Images for ASR and TTS through Single-loop and Dual-loop Multimodal Chain Framework
• Fine-Tuning StyleGAN2 For Cartoon Face Generation
• MaskCycleGAN-VC: Learning Non-parallel Voice Conversion with Filling in Frames
• Mutual-GAN: Towards Unsupervised Cross-Weather Adaptation with Mutual Information Constraint
• Unsupervised Many-to-Many Image-to-Image Translation Across Multiple Domains
• Supervised and Unsupervised Learning of Parameterized Color Enhancement
• Domain2Vec: Domain Embedding for Unsupervised Domain Adaptation
• k-Same-Siamese-GAN: k-Same Algorithm with Generative Adversarial Network for Facial Image De-identification with Hyperparameter Tuning and Mixed Precision Training
• Improving Shape Deformation in Unsupervised Image-to-Image Translation
• Crossing-Domain Generative Adversarial Networks for Unsupervised Multi-Domain Image-to-Image Translation
• Langevin Cooling for Domain Translation
• ReshapeGAN: Object Reshaping by Providing A Single Reference Image
• DSRGAN: Explicitly Learning Disentangled Representation of Underlying Structure and Rendering for Image Generation without Tuple Supervision
• Toward Learning a Unified Many-to-Many Mapping for Diverse Image Translation
• Unaligned Image-to-Sequence Transformation with Loop Consistency
• Learn to synthesize and synthesize to learn
• Distributed Generative Adversarial Net
• Sparse Pose Trajectory Completion
• TransferI2I: Transfer Learning for Image-to-Image Translation from Small Datasets
• TRIP: Refining Image-to-Image Translation via Rival Preferences
• Synthesizing Photorealistic Images with Deep Generative Learning
• Pose Augmentation: Class-agnostic Object Pose Transformation for Object Recognition
• On the Generative Utility of Cyclic Conditionals
• Smoothing the Disentangled Latent Style Space for Unsupervised Image-to-Image Translation
• Adversarial Transfer Learning for Cross-domain Visual Recognition
• When Autonomous Systems Meet Accuracy and Transferability through AI: A Survey
• Image-to-Image Translation of Synthetic Samples for Rare Classes
• Neural Wireframe Renderer: Learning Wireframe to Image Translations
• Unaligned Image-to-Image Translation by Learning to Reweight
• Expression Conditional GAN for Facial Expression-to-Expression Translation
• Video-to-Video Translation for Visual Speech Synthesis
• Mutual Information-based Disentangled Neural Networks for Classifying Unseen Categories in Different Domains: Application to Fetal Ultrasound Imaging
• Unsupervised Contrastive Photo-to-Caricature Translation based on Auto-distortion
• DGST : Discriminator Guided Scene Text detector
• Joint Wasserstein Distribution Matching
• Cross-Domain Latent Modulation for Variational Transfer Learning
• Systematic Analysis of Image Generation using GANs
• Training End-to-end Single Image Generators without GANs
• MirrorNet: A Deep Bayesian Approach to Reflective 2D Pose Estimation from Human Images
• Domain Adaptation with Morphologic Segmentation
• Face-to-Music Translation Using a Distance-Preserving Generative Adversarial Network with an Auxiliary Discriminator
• Learning to see across Domains and Modalities