- I have rearranged original patchcore codes (https://github.com/amazon-science/patchcore-inspection) in order to apply this State-of-the-Art approach more freely/lightly.
- Also I have tried to re-engineer it to make the model/approach applicable to edge device
- lighter feature extractor applicable to edge device, preferring MobileNet over WideResnet originally applied in patchcore. But some parameter tuning is required due to narrow channels of MobileNet.
- convert pytorch model to tflite model (8 bit model) via onnix converter.
- the relevant scripts are found in folder:
- '00_convert_middle_layer_feature_extractor_pytorch_model_to_tflite'
- I researched, in details step-by-step, how patchcore works
- see for scripts in folder '00_patchcore_details_check_step_by_step'
- When you need data augmentation
- see (for some help) scripts in folder '00_data_preparation_with_image_processing'
- 'main' folder: driver scripts
- 'src' and 'src2' folder: almost totally from original patchcore codes.
-
Notifications
You must be signed in to change notification settings - Fork 0
gouldberg/patchcore-MVTech-anomaly-detection
Folders and files
| Name | Name | Last commit message | Last commit date | |
|---|---|---|---|---|
Repository files navigation
About
No description, website, or topics provided.
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published