vandanavk / mxnet-gluon-gan

Highly scalable Apache MXNet based implementation of various Generative Adversarial Networks.

Apache MXNet Gluon - Generative Adversarial Networks (GAN)

Collection of Jupyter Notebooks with Apache MXNet GLUON implementation of various standard GAN papers.

Note: At this point of time, this repo may be used as reference implementations only. Hyperparameters, and various other factors are not fully tuned. Accuracies/any metric claimed in paper may NOT be exactly reflected with this implementations.

Prerequisites

pip install mxnet-mkl # For CPU machines
pip install mxnet-cu90mkl # For GPU machines

GAN Networks

No.	Title	Paper	Code
1	Generative Adversarial Networks (GAN)	https://arxiv.org/abs/1406.2661	GAN
2	Deep Convolutional GAN (DCGAN)	https://arxiv.org/abs/1511.06434	TODO
3	Conditional GAN (CGAN)	https://arxiv.org/abs/1411.1784	TODO
4	Pix2Pix	https://arxiv.org/abs/1611.07004	TODO
5	BEGAN: Boundary Equilibrium Generative Adversarial Networks	https://arxiv.org/abs/1703.10717	TODO
6	BicycleGAN: Toward Multimodal Image-to-Image Translation	https://arxiv.org/abs/1711.11586	TODO
7	Boundary seeking GAN	https://arxiv.org/abs/1611.06430	TODO
8	Context Conditional GAN	https://arxiv.org/abs/1611.06430	TODO
9	Context Encoders: Feature Learning by Inpainting	https://arxiv.org/abs/1604.07379	TODO
10	Coupled Generative Adversarial Networks	https://arxiv.org/abs/1606.07536	TODO
11	Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks	https://arxiv.org/abs/1703.10593	TODO
12	DiscoGAN - Learning to Discover Cross-Domain Relations with Generative Adversarial Networks	https://arxiv.org/abs/1703.05192	TODO
13	DualGAN: Unsupervised Dual Learning for Image-to-Image Translation	https://arxiv.org/abs/1703.05192	TODO
14	Energy-based Generative Adversarial Network	https://arxiv.org/abs/1609.03126	TODO
15	InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets	https://arxiv.org/abs/1606.03657	TODO
16	Unsupervised Pixel-Level Domain Adaptation with Generative Adversarial Networks	https://arxiv.org/abs/1612.05424	TODO
17	StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation	https://arxiv.org/abs/1711.09020	TODO
18	SRGAN: Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network	https://arxiv.org/abs/1609.04802	TODO
19	Improved Training of Wasserstein GANs	https://arxiv.org/abs/1704.00028	TODO

Contributions

All contributions welcome! You can contribute with bug fixes, changes, suggestions for new paper implementations. Please see above list of GAN networks I plan to implement, please feel free to create an issue and assign yourself any implementation.

See this issue with list of TODOs enhancements planned - sandeep-krishnamurthy#1

Credits, Motivation

1. This work is heavily motivated and influenced from this awesome project - https://github.com/eriklindernoren/keras-gan