README_en.md

Implementation of ALOCC, an anomary detection method with deep learning

Implemented with Chainer and PyTorch.

Sabokrou, et al. "Adversarially Learned One-Class Classifier for Novelty Detection", The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018, pp. 3379-3388

https://arxiv.org/abs/1802.09088

Requirements (PyTorch)

Requires PyTorch, PyTorch-Ignite, OpenCV, Matplotlib, and scikit-learn.

Installation:
PyTorch: see the official document.

sudo pip install pytorch-ignite opencv-python matplotlib sklearn

Run

Hyperparameters can be set by editting setting.json.

$ python train.py setting_file output_directory [-g GPUID]

Example:

$ python train.py setting.json result