README.md

AnoGAN2d

This is the implementation of "AnoGAN" corresponding to 2-dimensional shape.
Original paper: T. Schlegl, P. Seeböck, S. M. Waldstein, U. Schmidt-Erfurth, and G. Langs. Unsupervised anomaly detection with generative adversarial networks to guide marker discovery. In International Conference on Information Processing in Medical Imaging, 2017. link

Usage

1. Build

Please build the source file according to the procedure.

$ mkdir build
$ cd build
$ cmake ..
$ make -j4
$ cd ..

2. Dataset Setting

Recommendation

• MVTec Anomaly Detection Dataset (MVTec AD)
  This is a dataset for benchmarking anomaly detection methods with a focus on industrial inspection.
  Link: official

Setting

Please create a link for the dataset.
The following hierarchical relationships are recommended.

datasets
|--Dataset1
|    |--train
|    |    |--image1.png
|    |    |--image2.bmp
|    |    |--image3.jpg
|    |
|    |--valid
|    |--test_anomaly
|    |--test_normal
|
|--Dataset2
|--Dataset3

You should substitute the path of training normal data for "<training_path>", test anomaly data for "<test_anomaly_path>", test normal data for "<test_normal_path>", respectively.
The following is an example for "MVTecAD".

$ cd datasets
$ mkdir MVTecAD
$ cd MVTecAD
$ ln -s <training_path> ./train
$ ln -s <test_anomaly_path> ./test_anomaly
$ ln -s <test_normal_path> ./test_normal
$ cd ../..

3. Training

Setting

Please set the shell for executable file.

$ vi scripts/train.sh

The following is an example of the training phase.
If you want to view specific examples of command line arguments, please view "src/main.cpp" or add "--help" to the argument.

#!/bin/bash

DATA='MVTecAD'

./AnoGAN2d \
    --train true \
    --epochs 300 \
    --dataset ${DATA} \
    --size 256 \
    --batch_size 16 \
    --gpu_id 0 \
    --nc 3

Run

Please execute the following to start the program.

$ sh scripts/train.sh

4. Test

Setting

Please set the shell for executable file.

$ vi scripts/test.sh

The following is an example of the test phase.
If you want to view specific examples of command line arguments, please view "src/main.cpp" or add "--help" to the argument.

#!/bin/bash

DATA='MVTecAD'
HEATMAP_MAX=0.1

./AnoGAN2d \
    --test true \
    --dataset ${DATA} \
    --test_dir "test_anomaly" \
    --test_result_dir "test_result_anomaly" \
    --heatmap_max ${HEATMAP_MAX} \
    --size 256 \
    --gpu_id 0 \
    --nc 3
    
./AnoGAN2d \
    --test true \
    --dataset ${DATA} \
    --test_dir "test_normal" \
    --test_result_dir "test_result_normal" \
    --heatmap_max ${HEATMAP_MAX} \
    --size 256 \
    --gpu_id 0 \
    --nc 3

Run

Please execute the following to start the program.

$ sh scripts/test.sh

5. Anomaly Detection

Setting

Please set the shell for executable file.

$ vi scripts/anomaly_detection.sh

The following is an example of the anomaly detection phase.
If you want to view specific examples of command line arguments, please view "src/main.cpp" or add "--help" to the argument.

#!/bin/bash

DATA='MVTecAD'

./AnoGAN2d \
    --AD true \
    --dataset ${DATA} \
    --anomaly_path "test_result_anomaly/anomaly_score.txt" \
    --normal_path "test_result_normal/anomaly_score.txt" \
    --n_thresh 256

Run

Please execute the following to start the program.

$ sh scripts/anomaly_detection.sh

Acknowledgments

This code is inspired by pytorch_advanced.