Unsupervised Integrated-Circuit Defect Segmentation via Image-Intrinsic Normality
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912583044300800 |
|---|---|
| author | Zhao, Botong Shi, Qijun Lyu, Shujing Lu, Yue |
| author_facet | Zhao, Botong Shi, Qijun Lyu, Shujing Lu, Yue |
| contents | Modern Integrated-Circuit(IC) manufacturing introduces diverse, fine-grained defects that depress yield and reliability. Most industrial defect segmentation compares a test image against an external normal set, a strategy that is brittle for IC imagery where layouts vary across products and accurate alignment is difficult. We observe that defects are predominantly local, while each image still contains rich, repeatable normal patterns. We therefore propose an unsupervised IC defect segmentation framework that requires no external normal support. A learnable normal-information extractor aggregates representative normal features from the test image, and a coherence loss enforces their association with normal regions. Guided by these features, a decoder reconstructs only normal content; the reconstruction residual then segments defects. Pseudo-anomaly augmentation further stabilizes training. Experiments on datasets from three IC process stages show consistent improvements over existing approaches and strong robustness to product variability. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_09375 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Unsupervised Integrated-Circuit Defect Segmentation via Image-Intrinsic Normality Zhao, Botong Shi, Qijun Lyu, Shujing Lu, Yue Computer Vision and Pattern Recognition Modern Integrated-Circuit(IC) manufacturing introduces diverse, fine-grained defects that depress yield and reliability. Most industrial defect segmentation compares a test image against an external normal set, a strategy that is brittle for IC imagery where layouts vary across products and accurate alignment is difficult. We observe that defects are predominantly local, while each image still contains rich, repeatable normal patterns. We therefore propose an unsupervised IC defect segmentation framework that requires no external normal support. A learnable normal-information extractor aggregates representative normal features from the test image, and a coherence loss enforces their association with normal regions. Guided by these features, a decoder reconstructs only normal content; the reconstruction residual then segments defects. Pseudo-anomaly augmentation further stabilizes training. Experiments on datasets from three IC process stages show consistent improvements over existing approaches and strong robustness to product variability. |
| title | Unsupervised Integrated-Circuit Defect Segmentation via Image-Intrinsic Normality |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2509.09375 |