cpark321/anomaly-detection

anomaly-detection

• Anomaly detection

  1. vanilla CNN (supervised)
  2. resnet18 (supervised)

• Dependencies

  • Python 3.6+
  • PyTorch==1.3
  • Dataset: MVTec Anomaly Detection Dataset

Dataset structure

./data   
│
├── bottle
│   ├── ground_truth
│   ├── train
│   └── test
│
├── carpet
├── leather
├── grid

Train model

• Run the following command.

python train.py --target bottle --model resnet18 -c 0 --lr 0.001

Reference

1. bayesian-optimization python library