Pytorch_AnormalyDetection

• Pytorch
• Anomaly Detection
• Python

Version

• Python : 3.7.6
• torch : 1.8.0+cu111
• torchvision : 0.9.0+cu111
• cv2 : 4.4.0
• skimage : 0.14.2
• matplotlib : 3.0.3
• numpy : 1.18.2

Usage Example

Dataset

MvTEC Anormaly detection dataset.
Link : https://www.mvtec.com/company/research/datasets/mvtec-ad/

1. Open scripts/main.py file

2. Set Parameter for Train, Test

• learning_rate : optimizer learning late
• num_epoch : train epoch
• batch_size : batch size
• data_dir : train data directory
• test_dir : test data directory
• backup_dir : save model directory
• train_loop : True = Run train, test, False = only test
• use_ssim : True = use ssim, False = just diff (cv2.absdiff())
• diff_thresh : diff threshold value

3. Run main.py

4. Result

• use ssim

• just cv2.absdiff