/Pytorch-GANs

Pytorch implementation of Generative Adversarial Networks (GAN) for ULTRASOUND image.

Primary LanguageJupyter Notebook

Pytorch GANs for ultrasound images

PyTorch로 구현한 GAN(Generative Adversarial Networks)
초음파 영상 버전으로 확장하기 위한 前 단계(for study)

How to use

$ git clone https://github.com/Jihunlee326/Pytorch-GANs
$ cd models/(사용하고 싶은 모델 경로)/
$ python trainer.py

학습 파라미터는 trainer.py 에서 직접 관리합니다.
datasets 폴더에 학습 이미지를 저장하면 됩니다.
결과 영상은 models 폴더 내 images 폴더에 저장됩니다.

Development Environment

  • NVIDIA GeFore 940MX
  • cuda 8.0
  • python 3.6.5
  • pytorch 0.4.0
  • torchvision 0.2.1

Generative Adversarial Networks

Vanilla GAN

Generative Adversarial Network

[Paper] [Code]

CGAN

Conditional Generative Adversarial Network

[Paper] [Code]

DCGAN

Deep Convolutional Generative Adversarial Network

[Paper] [Code]

ACGAN

Auxiliary Classifier Generative Adversarial Network

[Paper] [Code]

InfoGAN

Information Maximizing Generative Adversarial Nets

[Paper] [Code]

Acknowledgements

This implementation has been based on this repository and tested with Pytorch over ver 0.4.0 on Windows 10 and Ubuntu 14.04.

To do list

학습 결과 이미지 첨부하기
논문 내용 요약하기