Adaptive Deviation Learning

Official Pytorch Implementation of Adaptive Deviation Learning (ADL)

Installation

Our results were computed using Python 3.8, with packages and respective version noted in requirements.txt. Please install the dependencies

pip install -r requirements.txt

Datasets

• To train on the MVtec Anomaly Detection dataset download Extract and move it to ./data/ folder.
• To train on the Visual Anomaly Dataset (VisA) download Extract and move it to ./data/ folder
• The Describable Textures dataset was used as the anomaly source. Extract and move it to ./data/ folder.

Training

Pass the folder containing the training dataset to the adl_train.py script as the --dataset_root argument and the folder locating the anomaly source images as the --anomaly_source_path argument. The training script also requires the epochs (--epochs), path to store checkpoint path (--experiment_dir), checkpoint weight name (--weight_name), contamination ratio (--cont). Example:

python adl_train.py --dataset_root=./data/mvtec_anomaly_detection --dataset=mvtec --anomaly_source_path ./data/dtd/images --classname=bottle --experiment_dir=./experiment --epochs=25 --cont=0.1 --weight_name=model_bal_10.pkl --report_name=result_report_mvtec_bal10

Evaluation

For Evaluation run eval_adl.py script with required arguments. Example:

python eval_adl.py --dataset_root=./data/mvtec_anomaly_detection --dataset=mvtec --anomaly_source_path ./data/dtd/images --classname=bottle --experiment_dir=./experiment --cont=0.1  --weight_name=model_bal_10.pkl --report_name=result_report_mvtec_bal10