Metraipolis › archive › dc_9091ef2810ea
Neural-Collapse-guided Task-Free Continual Anomaly Detection — arXiv:2609.03406
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.03406
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T02:56:17.714Z
- SHA-256 of the archived text
ab835911914596d5baab58d7d438ad63bb3690131ce681ef2ad3d7d2849e056d- Stable URL of this page
https://metraipolis.com/documento/dc_9091ef2810ea
How to cite
Neural-Collapse-guided Task-Free Continual Anomaly Detection — arXiv:2609.03406 — arXiv (en), https://arxiv.org/abs/2609.03406 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_9091ef2810ea (Metraipolis archive dc_9091ef2810ea, sha256 ab835911914596d5baab58d7d438ad63bb3690131ce681ef2ad3d7d2849e056d)
BibTeX
@misc{dc9091ef2810ea,
title = {Neural-Collapse-guided Task-Free Continual Anomaly Detection — arXiv:2609.03406},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.03406},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_9091ef2810ea; sha256 ab835911914596d5baab58d7d438ad63bb3690131ce681ef2ad3d7d2849e056d},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.03406. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Recent years have witnessed growing interest in continual anomaly detection for industrial visual inspection. However, real-world manufacturing environments exhibit unpredictable shifts in data distributions, rendering task-dependent continual learning assumptions impractical. To address this limitation, we formulate industrial anomaly detection as a task-free continual learning problem and propose NC-TFAD, a neural-collapse-inspired, geometry-driven framework for learning from non-stationary data streams without task boundaries. NC-TFAD freezes a pretrained backbone and aligns streaming features to a simplex Equiangular Tight Frame (ETF) prototype space to stabilize representation geometry under non-stationary streams. To satisfy the NC-inspired geometric construction in the absence of real anomalies, we generate synthetic anomaly samples as auxiliary anchors during training. Building on this geometry, we further introduce inter- and intra-class regularization together with a Focal Neural Collapse Contrastive (FNCC) loss to suppress representation drift and improve normal-anomaly separability. Finally, a normal-patch-prototype-guided localization branch constructs calibrated patch-wise deviation maps from normal training samples and fuses them with a weak self-attention prior, producing anomaly heatmaps without pixel-level annotations. Extensive experiments on MVTec AD and VisA show that NC-TFAD consistently outperforms representative task-free continual learning methods adapted from general vision, as well as unified anomaly detection baselines, in both image-level detection and pixel-level localization under the task-free continual learning protocol. These results highlight that geometry-driven modeling offers an effective and robust solution for task-free continual anomaly detection in real-world industrial applications. Tema degli esploratori: statistica e scienza dei dati (retro-catalogo arXiv: annunci del 2026-09-04 in cs.CV). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV). Pagina del lavoro: https://arxiv.org/abs/2609.03406 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T02:27:18.956Z dalla pagina https://arxiv.org/catchup/cs.CV/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).