This is a framework for anomaly detection mainly based on GAN and AE. This framework is designed to be hierarchical and visual. This framework can be easily adjusted to different anomaly detection methods such as anogan, ganomaly, etc.

Now the GANomaly and f-AnoGAN is accomplished.

For GANomaly, run:

python train.py dataset yourdataset --nepoch 100 --model ganomaly

For f-AnoGAN, run:

python train.py dataset yourdataset --nepoch 100 --nenepoch 100 --model fanogan

You can load pretrained model by add

--load_final_weigths and --load_final_en_weigths