The idea is to train deep neural networks with synthetic data only or in addition to regular training data to improve the net performance. This repository aims to give an overview of different techniques which can be useful for generating 2D training data.
The following papers are relevant in the area of data generation and it is likely that the generated data can be used for training. Papers are ordered by their submission date.
Inspired by the 3D-Machine-Learning overview by timzhang642, I started making this list while working on my master thesis.
We are on [matrix] to discuss, share knowledge and ask questions.
To contribute to the repository, you may add content through pull requests or open an issue.
- Data Augmentation
- Image Synthesis
- Domain Synthesis
- Texture Synthesis
- Image-to-Image Translation
- Style Transfer
- Overviews
- Regularization
- Training
- Evaluation
- Flip
- Rotation
- Scale
- Crop
- Translation
- Noise
- Gaussian
- On single image
- Apply textures
- Change lighting
- Change object details
- Add random objects
- Change viewpoint
- Change background
- On dataset
- Exchange segment patches within dataset
- Image synthesis
Data Augmentation using Random Image Cropping and Patching for Deep CNNs (2015) [Paper]
- An Introduction to Image Synthesis with Generative Adversarial Nets (2018)
- A Survey of Image Synthesis and Editing with Generative Adversarial Networks (2017)
Large Scale GAN Training for High Fidelity Natural Image Synthesis (2018) [Paper] [Code/Live Demo]
Self-Attention Generative Adversarial Networks (2018) [Paper]
Training Deep Networks with Synthetic Data: Bridging the Reality Gap by Domain Randomization (2018) [Paper]
Generative Semantic Manipulation with Contrasting GAN (2017) [Paper]
Synthesized Texture Quality Assessment via Multi-scale Spatial and Statistical Texture Attributes of Image and Gradient Magnitude Coefficients (2018) [Paper]
High-Resolution Multi-Scale Neural Texture Synthesis (2017) [Paper]
Texture Synthesis Using Convolutional Neural Networks (2015) [Paper]
Harmonic Unpaired Image-to-Image Translation (2019) [Paper]
Unsupervised Attention-guided Image-to-Image Translation (2018) [Paper] [Code (TensorFlow)] [Code (PyTorch)]
Show, Attend and Translate: Unsupervised Image Translation with Self-Regularization and Attention (2018) [Paper]
Exemplar Guided Unsupervised Image-to-Image Translation with Semantic Consistency (2018) [Paper]
Attention-GAN for Object Transfiguration in Wild Images (2018) [Paper]
Image to Image Translation for Domain Adaptation (2017) [Paper]
High-Resolution Image Synthesis and Semantic Manipulation with Conditional GANs (2017) [Website] [Paper] [Code] [Video]
Toward Multimodal Image-to-Image Translation (2017) [Website] [Paper] [Code (PyTorch)] [Code (TensorFlow)] [Video]
StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation (2017) [Paper]
Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks (2017) [Paper] [Code (PyTorch)] [Code (Torch)]
Learning to Discover Cross-Domain Relations with Generative Adversarial Networks (2017) [Paper]
DualGAN: Unsupervised Dual Learning for Image-to-Image Translation (2017) [Paper]
Image-to-Image Translation with Conditional Adversarial Nets (2017) [Website] [Paper] [Code] [Live Demo]
Arbitrary Style Transfer with Style-Attentional Networks (2018) [Paper]
Photo-realistic Facial Texture Transfer (2017) [Paper]
TextureGAN: Controlling Deep Image Synthesis with Texture Patches (2017) [Paper]
Exploring the structure of a real-time, arbitrary neural artistic stylization network (2017) [Paper] [Code] [Live Demo] [[Code (Live Demo)]](Arbitrary Style Transfer in the Browser)
Image Style Transfer Using Convolutional Neural Networks (2016) [Paper]
A Neural Algorithm of Artistic Style (2015) [Paper]
Generative Adversarial Networks: An Overview [Paper]
Improved Techniques for Training GANs (2016) [Paper]
- Inception Score — evaluating the realism of your GAN
- What is the rationale behind "Inception Score" as a metric for quality of generative models (e.g. GANs)?