PatchCore MVTec AD ONNX models

README.md

@@ -0,0 +1,67 @@
+---
+license: cc-by-nc-sa-4.0
+library_name: onnx
+pipeline_tag: image-classification
+tags:
+  - anomaly-detection
+  - patchcore
+  - mvtec-ad
+  - onnx
+  - industrial-inspection
+datasets:
+  - mvtec-ad
+---
+
+# PatchCore on MVTec AD (ONNX, runs in the browser)
+
+From-scratch PyTorch reimplementation of **PatchCore** (Roth et al., *Towards Total Recall in Industrial Anomaly
+Detection*, CVPR 2022), exported to ONNX. Fitted on **defect-free images only**; defects are found as patches that
+are far from every patch seen on good parts.
+
+**Live demo (in your browser, no server):** [yandyassine/patchcore-defect-detection](https://huggingface.co/spaces/yandyassine/patchcore-defect-detection) ·
+**Code & write-up:** [COMPUTER-VISION-PROJECT-LIBRARY / 13](https://github.com/yandouziyassine/COMPUTER-VISION-PROJECT-LIBRARY/tree/main/13-anomaly-detection-patchcore)
+
+## Results (MVTec AD test set, image level)
+
+| Category | Test images | Image AUROC | F1 at the shipped threshold |
+|---|---|---|---|
+| bottle | 83 | 1.000 | 1.000 |
+| metal_nut | 115 | 0.999 | 0.995 |
+| grid | 78 | 0.977 | 0.974 |
+
+These numbers were re-measured **with the ONNX files in this repo** (fp16-stored weights), and match the PyTorch
+model exactly at this precision.
+
+## Files
+
+| File | What it does | Size |
+|---|---|---|
+| `backbone.onnx` | image `(1,3,224,224)` → patch features `(784, 1536)`: WideResNet-50-2 layers 2+3, 3×3 local averaging | 50 MB |
+| `knn_<category>.onnx` | features → anomaly map `(224,224)` + image score; memory bank (10 % greedy coreset) baked in | 50–64 MB |
+| `thresholds.json` | per-category decision threshold (best F1 on the test set) | |
+
+Weights are stored as float16 and cast to float32 inside the graph, so the math runs in float32.
+
+## Use
+
+Preprocessing: resize the shorter side to 256, center-crop 224, ImageNet mean/std normalization.
+
+```python
+import numpy as np, onnxruntime as ort
+bb = ort.InferenceSession("backbone.onnx")
+knn = ort.InferenceSession("knn_bottle.onnx")
+feats = bb.run(None, {"image": x})[0]            # x: float32 (1,3,224,224)
+amap, score = knn.run(None, {"features": feats})
+defect = score > 1.9728                           # thresholds.json["bottle"]["threshold"]
+```
+
+## Limitations
+
+- One model per product; a photo of another object will look anomalous everywhere (by design).
+- Thresholds were picked on the MVTec test set itself, so the F1 column is optimistic; AUROC is threshold-free.
+- Trained at 224 px; very small defects can be missed.
+
+## License
+
+Memory banks are derived from MVTec AD, licensed **CC BY-NC-SA 4.0**: non-commercial use only. Backbone weights:
+torchvision ImageNet WideResNet-50-2.

backbone.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:220153c1693b2fc5a00b9f3261c35ab4ec3d66ac2e554d596407a8e83c867e5e
+size 49734896

knn_bottle.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:11683ee720d10143b3ecf9841f0e116eb1e0dfc8d0d9ea93edec08bc29539793
+size 50373353

knn_grid.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:223d64a726c168d1b2d56003b7150c1dd110a9ef92ea77e8d8f679734d58b0b9
+size 63628441

knn_metal_nut.onnx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32b3aa7b53bc043d9b70ccf90d7d5806c4c65646c0c15c7a3afaf1b5eebc1d1d
+size 53026215

thresholds.json

@@ -0,0 +1,20 @@
+{
+ "bottle": {
+  "threshold": 1.9728,
+  "f1": 1.0,
+  "precision": 1.0,
+  "recall": 1.0
+ },
+ "metal_nut": {
+  "threshold": 2.1994,
+  "f1": 0.9947,
+  "precision": 0.9894,
+  "recall": 1.0
+ },
+ "grid": {
+  "threshold": 1.7589,
+  "f1": 0.9735,
+  "precision": 0.9821,
+  "recall": 0.9649
+ }
+}