Experiment Results
March 1, 2022 ยท View on GitHub
To train self-supervised model we used same hyperparameters as was used in paper:
| Hyperparameter | Value |
|---|---|
| Number of epochs | 265 |
| Batch size | 32 |
| Learning rate | 0.03 |
| Input size | 256 |
AUC comparison of our code and paper results
| Defect Name | CutPaste binary (ours) | CutPaste binary (paper's) | CutPaste 3 way (ours) | CutPaste 3 way (paper's) |
|---|---|---|---|---|
| tile | 84.1 | 95.9 | 78.9 | 93.4 |
| wood | 89.5 | 94.9 | 89.2 | 98.6 |
| pill | 88.7 | 93.4 | 78.7 | 92.4 |
| leather | 98.7 | 99.7 | 84.8 | 100.0 |
| hazelnut | 98.8 | 91.3 | 80.8 | 97.3 |
| screw | 89.2 | 54.4 | 56.6 | 86.3 |
| cable | 83.3 | 87.7 | 75.7 | 93.1 |
| toothbrush | 94.7 | 99.2 | 78.6 | 98.3 |
| capsule | 80.2 | 87.9 | 70.8 | 96.2 |
| carpet | 57.9 | 67.9 | 26.1 | 93.1 |
| zipper | 99.5 | 99.4 | 85.7 | 99.4 |
| metal_nut | 91.5 | 96.8 | 89.7 | 99.3 |
| bottle | 98.5 | 99.2 | 75.7 | 98.3 |
| grid | 99.9 | 99.9 | 73.0 | 99.9 |
| transistor | 84.4 | 96.4 | 85.5 | 95.5 |
ROC curves using embeddings from binary classification for self-supervised learning
Click to see ROC curves!
Self-supervised binary training results
Click to see self-supervised training results!
Training accuracy and loss for bottle
Training accuracy and loss for pill
Training accuracy and loss for cable
Training accuracy and loss for capsule
Training accuracy and loss for tile
Self-supervised 3-way training results
Click to see self-supervised training results!
Training accuracy and loss for pill
Training accuracy and loss for screw
Training accuracy and loss for tile
Training accuracy and loss for zipper
t-SNE visualisation of embeddings
Click to see t-SNE visualisations!
