| query image | generated similar image | differece | |:---:|:---:|:---:| | | | | | | | | | | | | | | | | | | | | | | | |

AnoGAN keras implementation

Unsupervised anomaly detection with DCGAN

Requirements

• Python 3.6
• OpenCV 3.4.0 (option: build from src with highgui)
• h5py
• scikit-learn
• PyQt5
• tqdm
• Keras 2.1.4
• TensorFlow 1.5.0

Usage

First, check directory structure

├── main.py
├── anogan.py 
├── weights
    ├── discriminator.h5
    └── generator.h5
└── result
    └── save the generated images when training

To test this project

$ python main.py

To train a model

$ python main.py --mode train

Then, the training steps(image) will be saved 'result' directory

usage: main.py [-h] [--img_idx IMG_IDX] 
                    [--label_idx LABEL_IDX] 
                    [--mode MODE]

Reference

paper : https://arxiv.org/abs/1703.05921
AnoGAN(code, keras) : https://github.com/yjucho1/anoGAN
AnoGAN(code, tf) : https://github.com/LeeDoYup/AnoGAN