Unsupervised Image-to-Image Translation Networks
arXiv:1703.00848
Abstract
Unsupervised image-to-image translation aims at learning a joint distribution of images in different domains by using images from the marginal distributions in individual domains. Since there exists an infinite set of joint distributions that can arrive the given marginal distributions, one could infer nothing about the joint distribution from the marginal distributions without additional assumptions. To address the problem, we make a shared-latent space assumption and propose an unsupervised image-to-image translation framework based on Coupled GANs. We compare the proposed framework with competing approaches and present high quality image translation results on various challenging unsupervised image translation tasks, including street scene image translation, animal image translation, and face image translation. We also apply the proposed framework to domain adaptation and achieve state-of-the-art performance on benchmark datasets. Code and additional results are available in https://github.com/mingyuliutw/unit .
NIPS 2017, 11 pages, 6 figures
References in corpus (4)
Cited by in corpus (106)
- Style Transfer from Non-Parallel Text by Cross-Alignment
- DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation
- Augmented CycleGAN: Learning Many-to-Many Mappings from Unpaired Data
- Deep Appearance Models for Face Rendering
- Multimodal Unsupervised Image-to-Image Translation
- A Review on Generative Adversarial Networks: Algorithms, Theory, and Applications
- Unsupervised Machine Translation Using Monolingual Corpora Only
- DRIT++: Diverse Image-to-Image Translation via Disentangled Representations
- An Introduction to Image Synthesis with Generative Adversarial Nets
- StressGAN: A Generative Deep Learning Model for 2D Stress Distribution Prediction
- Video-to-Video Synthesis
- Train Sparsely, Generate Densely: Memory-efficient Unsupervised Training of High-resolution Temporal GAN
- One-Shot Unsupervised Cross Domain Translation
- CariGANs: Unpaired Photo-to-Caricature Translation
- Deep Active Inference
- Triangle Generative Adversarial Networks
- UNIT-DDPM: UNpaired Image Translation with Denoising Diffusion Probabilistic Models
- Domain Stylization: A Strong, Simple Baseline for Synthetic to Real Image Domain Adaptation
- Hypernetwork functional image representation
- DeepRoad: GAN-based Metamorphic Autonomous Driving System Testing
- Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation
- A Closed-form Solution to Photorealistic Image Stylization
- ControlVAE: Controllable Variational Autoencoder
- Virtual Mixup Training for Unsupervised Domain Adaptation
- A Universal Music Translation Network
- GANimation: Anatomically-aware Facial Animation from a Single Image
- Image Captioning with Very Scarce Supervised Data: Adversarial Semi-Supervised Learning Approach
- Scene Text Synthesis for Efficient and Effective Deep Network Training
- Semantically Decomposing the Latent Spaces of Generative Adversarial Networks
- An Unsupervised Domain Adaptation Model based on Dual-module Adversarial Training
- Mask-aware Photorealistic Face Attribute Manipulation
- ELEGANT: Exchanging Latent Encodings with GAN for Transferring Multiple Face Attributes
- AUTO3D: Novel view synthesis through unsupervisely learned variational viewpoint and global 3D representation
- SPIGAN: Privileged Adversarial Learning from Simulation
- Unpaired Speech Enhancement by Acoustic and Adversarial Supervision for Speech Recognition
- Unsupervised Cipher Cracking Using Discrete GANs
- Text-to-Image-to-Text Translation using Cycle Consistent Adversarial Networks
- DCAN: Dual Channel-wise Alignment Networks for Unsupervised Scene Adaptation
- Open Set Domain Adaptation by Backpropagation
- Causal Generative Domain Adaptation Networks
- Geometry-Consistent Generative Adversarial Networks for One-Sided Unsupervised Domain Mapping
- Twin-GAN -- Unpaired Cross-Domain Image Translation with Weight-Sharing GANs
- Deep Adversarial Attention Alignment for Unsupervised Domain Adaptation: the Benefit of Target Expectation Maximization
- Unsupervised Video-to-Video Translation
- Latent Translation: Crossing Modalities by Bridging Generative Models
- Human Action Generation with Generative Adversarial Networks
- Label-Driven Reconstruction for Domain Adaptation in Semantic Segmentation
- T2Net: Synthetic-to-Realistic Translation for Solving Single-Image Depth Estimation Tasks
- Neural Rendering and Reenactment of Human Actor Videos
- Constraint-Based Visual Generation
- Multi-View Data Generation Without View Supervision
- DART: Domain-Adversarial Residual-Transfer Networks for Unsupervised Cross-Domain Image Classification
- Few-shot Video-to-Video Synthesis
- Recursive Chaining of Reversible Image-to-image Translators For Face Aging
- 3D Hand Pose Estimation using Simulation and Partial-Supervision with a Shared Latent Space
- Learning from Multi-domain Artistic Images for Arbitrary Style Transfer
- GANHopper: Multi-Hop GAN for Unsupervised Image-to-Image Translation
- Instance-level Facial Attributes Transfer with Geometry-Aware Flow
- A Unified Framework for Generalizable Style Transfer: Style and Content Separation
- SDIT: Scalable and Diverse Cross-domain Image Translation
- Deep learning approach in multi-scale prediction of turbulent mixing-layer
- Cycle-Consistent Adversarial Learning as Approximate Bayesian Inference
- toon2real: Translating Cartoon Images to Realistic Images
- WarpGAN: Automatic Caricature Generation
- Polite Dialogue Generation Without Parallel Data
- CNNs and GANs in MRI-based cross-modality medical image estimation
- Learning to Read by Spelling: Towards Unsupervised Text Recognition
- Deep Video Portraits
- Generative Adversarial Network for Image Synthesis
- Disentanglement for Discriminative Visual Recognition
- Resembled Generative Adversarial Networks: Two Domains with Similar Attributes
- Play as You Like: Timbre-enhanced Multi-modal Music Style Transfer
- GAN-Based Facial Attractiveness Enhancement
- Domain-Specific Mappings for Generative Adversarial Style Transfer
- Cross Domain Image Generation through Latent Space Exploration with Adversarial Loss
- TET-GAN: Text Effects Transfer via Stylization and Destylization
- GANtruth - an unpaired image-to-image translation method for driving scenarios
- Joint haze image synthesis and dehazing with mmd-vae losses
- Unsupervised Stylish Image Description Generation via Domain Layer Norm
- Composable Unpaired Image to Image Translation
- Learning Discriminators as Energy Networks in Adversarial Learning
- Learning image from projection: a full-automatic reconstruction (FAR) net for sparse-views computed tomography
- Guiding the One-to-one Mapping in CycleGAN via Optimal Transport
- Unsupervised Latent Space Translation Network
- Intrinsic Autoencoders for Joint Neural Rendering and Intrinsic Image Decomposition
- PortraitGAN for Flexible Portrait Manipulation
- Sim2Real for Self-Supervised Monocular Depth and Segmentation
- Generative Creativity: Adversarial Learning for Bionic Design
- Memory-guided Unsupervised Image-to-image Translation
- 3D Guided Fine-Grained Face Manipulation
- Few-Shot Unsupervised Image-to-Image Translation on complex scenes
- Generative Adversarial Networks for Video-to-Video Domain Adaptation
- Mutation Testing framework for Machine Learning
- Variational learning across domains with triplet information
- MCMI: Multi-Cycle Image Translation with Mutual Information Constraints
- Learning Task-oriented Disentangled Representations for Unsupervised Domain Adaptation
- G2C: A Generator-to-Classifier Framework Integrating Multi-Stained Visual Cues for Pathological Glomerulus Classification
- Video-to-Video Translation for Visual Speech Synthesis
- Bridging the Gap between Label- and Reference-based Synthesis in Multi-attribute Image-to-Image Translation
- Unsupervised Meta-learning of Figure-Ground Segmentation via Imitating Visual Effects
- Unsupervised Image-to-Image Translation with Stacked Cycle-Consistent Adversarial Networks
- Synthesizing Photorealistic Images with Deep Generative Learning
- Deep Factorised Inverse-Sketching
- PerformanceNet: Score-to-Audio Music Generation with Multi-Band Convolutional Residual Network
- SMET: Scenario-based Metamorphic Testing for Autonomous Driving Models
- Head2HeadFS: Video-based Head Reenactment with Few-shot Learning