IAD-Unify: A Region-Grounded Unified Model for Industrial Anomaly Segmentation, Understanding, and Generation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914471122829312 |
|---|---|
| author | Zheng, Haoyu Lin, Tianwei Wang, Wei Wang, Zhuonan Zhang, Wenqiao Zhu, Jiaqi Shao, Feifei |
| author_facet | Zheng, Haoyu Lin, Tianwei Wang, Wei Wang, Zhuonan Zhang, Wenqiao Zhu, Jiaqi Shao, Feifei |
| contents | Real-world industrial inspection requires not only localizing defects, but also explaining them in natural language and generating controlled defect edits. However, existing approaches fail to jointly support all three capabilities within a unified framework and evaluation protocol. We propose IAD-Unify, a dual-encoder unified framework in which a frozen DINOv2-based region expert supplies precise anomaly evidence to a shared Qwen3.5-4B vision-language backbone via lightweight token injection, jointly enabling anomaly segmentation, region-grounded understanding, and mask-guided generation. To enable unified evaluation, we further construct Anomaly-56K, a comprehensive unified multi-task IAD evaluation platform, spanning 59,916 images across 24 categories and 104 defect variants. Controlled ablations yield four findings: (i) region grounding is the decisive mechanism for understanding, removing it degrades location accuracy by >76 pp; (ii) predicted-region performance closely matches oracle, confirming deployment viability; (iii) region-grounded generation achieves the best full-image fidelity and masked-region perceptual quality; and (iv) pre-initialized joint training improves understanding at negligible generation cost (-0.16 dB). IAD-Unify further achieves strong performance on the MMAD benchmark, including categories unseen during training, demonstrating robust cross-category generalization. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_12440 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | IAD-Unify: A Region-Grounded Unified Model for Industrial Anomaly Segmentation, Understanding, and Generation Zheng, Haoyu Lin, Tianwei Wang, Wei Wang, Zhuonan Zhang, Wenqiao Zhu, Jiaqi Shao, Feifei Computer Vision and Pattern Recognition Artificial Intelligence Real-world industrial inspection requires not only localizing defects, but also explaining them in natural language and generating controlled defect edits. However, existing approaches fail to jointly support all three capabilities within a unified framework and evaluation protocol. We propose IAD-Unify, a dual-encoder unified framework in which a frozen DINOv2-based region expert supplies precise anomaly evidence to a shared Qwen3.5-4B vision-language backbone via lightweight token injection, jointly enabling anomaly segmentation, region-grounded understanding, and mask-guided generation. To enable unified evaluation, we further construct Anomaly-56K, a comprehensive unified multi-task IAD evaluation platform, spanning 59,916 images across 24 categories and 104 defect variants. Controlled ablations yield four findings: (i) region grounding is the decisive mechanism for understanding, removing it degrades location accuracy by >76 pp; (ii) predicted-region performance closely matches oracle, confirming deployment viability; (iii) region-grounded generation achieves the best full-image fidelity and masked-region perceptual quality; and (iv) pre-initialized joint training improves understanding at negligible generation cost (-0.16 dB). IAD-Unify further achieves strong performance on the MMAD benchmark, including categories unseen during training, demonstrating robust cross-category generalization. |
| title | IAD-Unify: A Region-Grounded Unified Model for Industrial Anomaly Segmentation, Understanding, and Generation |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2604.12440 |