{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,20]],"date-time":"2026-03-20T11:03:18Z","timestamp":1774004598057,"version":"3.50.1"},"reference-count":34,"publisher":"Oxford University Press (OUP)","issue":"3","license":[{"start":{"date-parts":[[2026,2,28]],"date-time":"2026-02-28T00:00:00Z","timestamp":1772236800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by-nc\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"published-print":{"date-parts":[[2026,2,28]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>Multi-modal industrial anomaly detection (IAD), which integrates RGB and 3D information, has become one of the key technical directions for improving detection robustness and accuracy. Although prevailing cross-modal feature-mapping methods are efficient and lightweight, they still suffer from two major limitations. First, they typically adopt a one-way modelling paradigm that regresses one modality from another and lack explicit interaction within a unified representation space, making it difficult to detect local, small-magnitude anomalies that appear only in a single modality. Second, fusion\u2013reconstruction methods derived from this paradigm rely on a single fusion stream optimized with a reconstruction loss. When trained solely on normal samples, this design can overgeneralize and lacks a parallel branch to enforce consistency constraints on the fused representations, which in turn limits reliable discrimination between normal and anomalous patterns in complex multi-modal scenarios. To address these issues, we propose FMFR, a feature-level multi-stage fusion and remapping framework that jointly models multi-stage feature fusion and cross-modal remapping. The framework consists of a fusion\u2013reconstruction branch and a remapping\u2013fusion branch, which are jointly constrained by a multi-order consistency loss. In the fusion\u2013reconstruction branch, a reconstruction loss supervises the intermediate fusion layers, encouraging them to learn joint representations that retain complete information and to reconstruct features without losing critical details. In the remapping\u2013fusion branch, the network learns bidirectional mappings between modalities and refuses the remapped features, while the multi-order consistency loss is used to align its fused representations with those of the fusion\u2013reconstruction branch. During inference, FMFR jointly leverages intra-modal reconstruction residuals, cross-modal remapping residuals, and the consistency deviation between the fused embeddings of the two branches to construct multi-source anomaly maps. This design forces anomalies to simultaneously violate both intra- and cross-modal priors, thereby suppressing the overgeneralization of a single fusion stream and enhancing the visibility of local anomaly structures that exist only in a single modality as well as the overall robustness of anomaly detection. Experimental results on the MVTec 3D-AD data set demonstrate that FMFR achieves competitive state-of-the-art performance on both anomaly detection and anomaly segmentation tasks.<\/jats:p>","DOI":"10.1093\/jcde\/qwag016","type":"journal-article","created":{"date-parts":[[2026,2,27]],"date-time":"2026-02-27T12:43:32Z","timestamp":1772196212000},"page":"233-245","source":"Crossref","is-referenced-by-count":0,"title":["FMFR: Feature-level multi-stage fusion and remapping for multi-modal industrial anomaly detection"],"prefix":"10.1093","volume":"13","author":[{"ORCID":"https:\/\/orcid.org\/0009-0000-3034-8081","authenticated-orcid":false,"given":"Chunshui","family":"Wang","sequence":"first","affiliation":[{"name":"School of Computer and Information Science, School of Software, Southwest University , Chongqing 400715 ,","place":["China"]}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5171-7835","authenticated-orcid":false,"given":"Heng","family":"Zhang","sequence":"additional","affiliation":[{"name":"School of Computer and Information Science, School of Software, Southwest University , Chongqing 400715 ,","place":["China"]}]}],"member":"286","published-online":{"date-parts":[[2026,2,28]]},"reference":[{"key":"2026032004512995600_bib1","doi-asserted-by":"crossref","first-page":"1135","DOI":"10.1007\/s12553-025-01011-4","article-title":"Optimized and interpretable machine learning framework for early breast cancer detection","volume":"15","author":"Ansari","year":"2025","journal-title":"Health and Technology"},{"key":"2026032004512995600_bib2","doi-asserted-by":"publisher","first-page":"75","DOI":"10.1007\/s44163-025-00307-8","article-title":"The role of explainable ai in enhancing breast cancer diagnosis using machine learning and deep learning models","volume":"5","author":"Ansari","year":"2025","journal-title":"Discover Artificial Intelligence"},{"key":"2026032004512995600_bib3","article-title":"The mvtec 3d-ad dataset for unsupervised 3d anomaly detection and localization","author":"Bergmann","year":"2021"},{"key":"2026032004512995600_bib4","doi-asserted-by":"crossref","first-page":"767","DOI":"10.1117\/12.3033181","article-title":"Dual-branch reconstruction network for industrial anomaly detection with rgb-d data","volume-title":"International Conference on Image, Signal Processing, and Pattern Recognition (ISPP 2024)","author":"Bi","year":"2024"},{"key":"2026032004512995600_bib5","first-page":"3586","article-title":"The eyecandies dataset for unsupervised multimodal anomaly detection and localization","volume-title":"Proceedings of the Asian Conference on Computer Vision","author":"Bonfiglioli","year":"2022"},{"key":"2026032004512995600_bib6","first-page":"9650","article-title":"Emerging properties in self-supervised vision transformers","volume-title":"Proceedings of the IEEE\/CVF international conference on computer vision","author":"Caron","year":"2021"},{"key":"2026032004512995600_bib7","doi-asserted-by":"crossref","first-page":"7038","DOI":"10.1145\/3581783.3611876","article-title":"Easynet: An easy network for 3d industrial anomaly detection","volume-title":"Proceedings of the 31st ACM International Conference on Multimedia","author":"Chen","year":"2023"},{"key":"2026032004512995600_bib8","doi-asserted-by":"crossref","first-page":"5000","DOI":"10.1109\/TII.2025.3552723","article-title":"Multimodal industrial anomaly detection via uni-modal and cross-modal fusion","volume":"21","author":"Cheng","year":"2025","journal-title":"IEEE Transactions on Industrial Informatics"},{"key":"2026032004512995600_bib9","first-page":"6185","article-title":"Shape-guided dual-memory learning for 3d anomaly detection","volume-title":"Proceedings of the 40th International Conference on Machine Learning","author":"Chu","year":"2023"},{"key":"2026032004512995600_bib10","first-page":"17234","article-title":"Multimodal industrial anomaly detection by crossmodal feature mapping","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Costanzino","year":"2024"},{"key":"2026032004512995600_bib11","first-page":"8445","article-title":"Rethinking reverse distillation for multi-modal anomaly detection","volume-title":"Proceedings of the AAAI Conference on Artificial Intelligence","author":"Gu","year":"2024"},{"key":"2026032004512995600_bib12","first-page":"98","article-title":"Cflow-ad: Real-time unsupervised anomaly detection with localization via conditional normalizing flows","volume-title":"Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision","author":"Gudovskiy","year":"2022"},{"key":"2026032004512995600_bib13","doi-asserted-by":"publisher","first-page":"41","DOI":"10.1093\/jcde\/qwaf043","article-title":"Unsupervised industrial anomaly detection using paired well-lit and low-light images","volume":"12","author":"Hoang","year":"2025","journal-title":"Journal of Computational Design and Engineering"},{"key":"2026032004512995600_bib14","doi-asserted-by":"publisher","first-page":"102823","DOI":"10.1016\/j.aei.2024.102823","article-title":"Daup: Enhancing point cloud homogeneity for 3d industrial anomaly detection via density-aware point cloud upsampling","volume":"62","author":"Li","year":"2024","journal-title":"Advanced Engineering Informatics"},{"key":"2026032004512995600_bib15","first-page":"328","article-title":"Learning diffusion models for multi-view anomaly detection","volume-title":"European Conference on Computer Vision","author":"Liu","year":"2024"},{"key":"2026032004512995600_bib16","article-title":"Multimodal industrial anomaly detection by crossmodal reverse distillation","author":"Liu","year":"2024"},{"key":"2026032004512995600_bib17","article-title":"Inp-former++: Advancing universal anomaly detection via intrinsic normal prototypes and residual learning","author":"Luo","year":"2025"},{"key":"2026032004512995600_bib18","doi-asserted-by":"publisher","first-page":"2440001","DOI":"10.1142\/S2811032324400010","article-title":"Masked autoencoders for 3d point cloud self-supervised learning","volume":"1","author":"Pang","year":"2023","journal-title":"World Scientific Annual Review of Artificial Intelligence"},{"key":"2026032004512995600_bib19","doi-asserted-by":"publisher","first-page":"2466","DOI":"10.1093\/jcde\/qwac115","article-title":"Novel industrial surface-defect detection using deep nested convolutional network with attention and guidance modules","volume":"9","author":"Park","year":"2022","journal-title":"Journal of Computational Design and Engineering"},{"key":"2026032004512995600_bib20","doi-asserted-by":"publisher","first-page":"46","DOI":"10.1093\/jcde\/qwaf085","article-title":"A robust anomaly detector for imbalanced industrial internet of things data","volume":"12","author":"Riaz","year":"2025","journal-title":"Journal of Computational Design and Engineering"},{"key":"2026032004512995600_bib21","first-page":"1907","article-title":"Same same but differnet: Semi-supervised defect detection with normalizing flows","volume-title":"Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision","author":"Rudolph","year":"2021"},{"key":"2026032004512995600_bib22","first-page":"1088","article-title":"Fully convolutional cross-scale-flows for image-based defect detection","volume-title":"Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision","author":"Rudolph","year":"2022"},{"key":"2026032004512995600_bib23","first-page":"2592","article-title":"Asymmetric student-teacher networks for industrial anomaly detection","volume-title":"Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision","author":"Rudolph","year":"2023"},{"key":"2026032004512995600_bib24","doi-asserted-by":"publisher","first-page":"103240","DOI":"10.1016\/j.aei.2025.103240","article-title":"Cpir: Multimodal industrial anomaly detection via latent bridged cross-modal prediction and intra-modal reconstruction","volume":"65","author":"Shangguan","year":"2025","journal-title":"Advanced Engineering Informatics"},{"key":"2026032004512995600_bib25","first-page":"24511","article-title":"Revisiting reverse distillation for anomaly detection","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Tien","year":"2023"},{"key":"2026032004512995600_bib26","first-page":"75","article-title":"Self-supervised feature adaptation for 3d industrial anomaly detection","volume-title":"European Conference on Computer Vision","author":"Tu","year":"2024"},{"key":"2026032004512995600_bib27","doi-asserted-by":"publisher","first-page":"326","DOI":"10.1093\/jcde\/qwae019","article-title":"Industrial defective chips detection using deep convolutional neural network with inverse feature matching mechanism","volume":"11","author":"Ullah","year":"2024","journal-title":"Journal of Computational Design and Engineering"},{"key":"2026032004512995600_bib28","doi-asserted-by":"publisher","first-page":"127483","DOI":"10.1016\/j.neucom.2024.127483","article-title":"Incremental template neighborhood matching for 3d anomaly detection","volume":"581","author":"Wang","year":"2024","journal-title":"Neurocomputing"},{"key":"2026032004512995600_bib29","doi-asserted-by":"publisher","first-page":"130073","DOI":"10.1016\/j.neucom.2025.130073","article-title":"Fusion-restoration model for industrial multimodal anomaly detection","volume":"637","author":"Wang","year":"2025","journal-title":"Neurocomputing"},{"key":"2026032004512995600_bib30","first-page":"8032","article-title":"Multimodal industrial anomaly detection via hybrid fusion","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Wang","year":"2023"},{"key":"2026032004512995600_bib31","article-title":"Dfr: Deep feature reconstruction for unsupervised anomaly segmentation","author":"Yang","year":"2020"},{"key":"2026032004512995600_bib32","article-title":"Fastflow: Unsupervised anomaly detection and localization via 2d normalizing flows","author":"Yu","year":"2021"},{"key":"2026032004512995600_bib33","first-page":"8330","article-title":"Draem-a discriminatively trained reconstruction embedding for surface anomaly detection","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision","author":"Zavrtanik","year":"2021"},{"key":"2026032004512995600_bib34","first-page":"2164","article-title":"Cheating depth: Enhancing 3d surface anomaly detection via depth simulation","volume-title":"Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision","author":"Zavrtanik","year":"2024"}],"container-title":["Journal of Computational Design and Engineering"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/academic.oup.com\/jcde\/advance-article-pdf\/doi\/10.1093\/jcde\/qwag016\/67174668\/qwag016.pdf","content-type":"application\/pdf","content-version":"am","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jcde\/article-pdf\/13\/3\/233\/67174668\/qwag016.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"syndication"},{"URL":"https:\/\/academic.oup.com\/jcde\/article-pdf\/13\/3\/233\/67174668\/qwag016.pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,20]],"date-time":"2026-03-20T08:51:43Z","timestamp":1773996703000},"score":1,"resource":{"primary":{"URL":"https:\/\/academic.oup.com\/jcde\/article\/13\/3\/233\/8502145"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,2,28]]},"references-count":34,"journal-issue":{"issue":"3","published-print":{"date-parts":[[2026,2,28]]}},"URL":"https:\/\/doi.org\/10.1093\/jcde\/qwag016","relation":{},"ISSN":["2288-5048"],"issn-type":[{"value":"2288-5048","type":"electronic"}],"subject":[],"published-other":{"date-parts":[[2026,3]]},"published":{"date-parts":[[2026,2,28]]}}}