Skip-GANomaly

This repository contains PyTorch implementation of the following paper: Skip-GANomaly: Skip Connected and Adversarially Trained Encoder-Decoder Anomaly Detection [1]

1. Table of Contents

• Skip-GANomaly
  • 1. Table of Contents
  • 2. Installation
  • 3. Experiment
  • 4. Training
    • 4.1. Train on CIFAR10
    • 4.2. Train on Custom Dataset
  • 5. Citing Skip-GANomaly
  • 6. Reference

2. Installation

1. First clone the repository

   git clone https://github.com/samet-akcay/skip-ganomaly.git
   

2. Create the virtual environment via conda

   conda create -n skipganomaly python=3.7
   

3. Activate the virtual environment.

   conda activate skipganomaly
   

4. Install the dependencies.

   pip install --user --requirement requirements.txt
   

3. Experiment

To replicate the results in the paper for CIFAR10 dataset, run the following commands:

# CIFAR
sh experiments/run_cifar.sh

4. Training

To list the arguments, run the following command:

python train.py -h

4.1. Training on CIFAR10

To train the model on CIFAR10 dataset for a given anomaly class, run the following:

python train.py \
    --dataset cifar10                                                             \
    --niter <number-of-epochs>                                                    \
    --abnormal_class                                                              \
        <airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck>    \
    --display                                   # optional if you want to visualize        

4.2. Train on Custom Dataset

To train the model on a custom dataset, the dataset should be copied into ./data directory, and should have the following directory & file structure:

Custom Dataset
├── test
│   ├── 0.normal
│   │   └── normal_tst_img_0.png
│   │   └── normal_tst_img_1.png
│   │   ...
│   │   └── normal_tst_img_n.png
│   ├── 1.abnormal
│   │   └── abnormal_tst_img_0.png
│   │   └── abnormal_tst_img_1.png
│   │   ...
│   │   └── abnormal_tst_img_m.png
├── train
│   ├── 0.normal
│   │   └── normal_tst_img_0.png
│   │   └── normal_tst_img_1.png
│   │   ...
│   │   └── normal_tst_img_t.png

Then model training is the same as the training explained above.

python train.py                     \
    --dataset <name-of-the-data>    \
    --isize <image-size>            \
    --niter <number-of-epochs>      \
    --display                       # optional if you want to visualize

For more training options, run python train.py -h.

5. Citing Skip-GANomaly

If you use this repository or would like to refer the paper, please use the following BibTeX entry

@INPROCEEDINGS {Akcay2019SkipGANomaly,
    author    = "Samet Akçay and Amir Atapour-Abarghouei andToby P. Breckon",
    title     = "Skip-GANomaly: Skip Connected and Adversarially Trained Encoder-Decoder Anomaly Detection",
    booktitle = "2019 International Joint Conference on Neural Networks (IJCNN)",
    year      = "2019",
    pages     = "1-8",
    month     = "jul",
    publisher = "IEEE"
}

6. Reference

[1] S. Akçay, A. Atapour-Abarghouei, T. P. Breckon, Skip-GANomaly: Skip Connected and Adversarially Trained Encoder-Decoder Anomaly Detection, in: International Joint Conference on Neural Networks (IJCNN), IEEE, 2019.