{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,17]],"date-time":"2026-03-17T18:39:00Z","timestamp":1773772740392,"version":"3.50.1"},"reference-count":35,"publisher":"MDPI AG","issue":"19","license":[{"start":{"date-parts":[[2022,10,10]],"date-time":"2022-10-10T00:00:00Z","timestamp":1665360000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100002347","name":"Federal Ministry of Education and Research (BMBF)","doi-asserted-by":"publisher","award":["01IS19083A"],"award-info":[{"award-number":["01IS19083A"]}],"id":[{"id":"10.13039\/501100002347","id-type":"DOI","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Sensors"],"abstract":"<jats:p>Image novelty detection is a repeating task in computer vision and describes the detection of anomalous images based on a training dataset consisting solely of normal reference data. It has been found that, in particular, neural networks are well-suited for the task. Our approach first transforms the training and test images into ensembles of patches, which enables the assessment of mean-shifts between normal data and outliers. As mean-shifts are only detectable when the outlier ensemble and inlier distribution are spatially separate from each other, a rich feature space, such as a pre-trained neural network, needs to be chosen to represent the extracted patches. For mean-shift estimation, the Hotelling T2 test is used. The size of the patches turned out to be a crucial hyperparameter that needs additional domain knowledge about the spatial size of the expected anomalies (local vs. global). This also affects model selection and the chosen feature space, as commonly used Convolutional Neural Networks or Vision Image Transformers have very different receptive field sizes. To showcase the state-of-the-art capabilities of our approach, we compare results with classical and deep learning methods on the popular dataset CIFAR-10, and demonstrate its real-world applicability in a large-scale industrial inspection scenario using the MVTec dataset. Because of the inexpensive design, our method can be implemented by a single additional 2D-convolution and pooling layer and allows particularly fast prediction times while being very data-efficient.<\/jats:p>","DOI":"10.3390\/s22197674","type":"journal-article","created":{"date-parts":[[2022,10,11]],"date-time":"2022-10-11T00:50:01Z","timestamp":1665449401000},"page":"7674","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":6,"title":["Fast and Efficient Image Novelty Detection Based on Mean-Shifts"],"prefix":"10.3390","volume":"22","author":[{"given":"Matthias","family":"Hermann","sequence":"first","affiliation":[{"name":"Institute for Optical Systems, HTWG Konstanz\u2014University of Applied Sciences, Alfred-Wachtel-Stra\u00dfe 8, 78462 Konstanz, Germany"}]},{"given":"Georg","family":"Umlauf","sequence":"additional","affiliation":[{"name":"Institute for Optical Systems, HTWG Konstanz\u2014University of Applied Sciences, Alfred-Wachtel-Stra\u00dfe 8, 78462 Konstanz, Germany"}]},{"given":"Bastian","family":"Goldl\u00fccke","sequence":"additional","affiliation":[{"name":"Department of Computer Science, University of Konstanz, Universit\u00e4tsstra\u00dfe 10, 78464 Konstanz, Germany"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-3789-8849","authenticated-orcid":false,"given":"Matthias O.","family":"Franz","sequence":"additional","affiliation":[{"name":"Institute for Optical Systems, HTWG Konstanz\u2014University of Applied Sciences, Alfred-Wachtel-Stra\u00dfe 8, 78462 Konstanz, Germany"}]}],"member":"1968","published-online":{"date-parts":[[2022,10,10]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"756","DOI":"10.1109\/JPROC.2021.3052449","article-title":"A unifying review of deep and shallow anomaly detection","volume":"109","author":"Ruff","year":"2021","journal-title":"Proc. IEEE"},{"key":"ref_2","doi-asserted-by":"crossref","unstructured":"Hermann, M., Goldl\u00fccke, B., and Franz, M.O. (2022, January 23\u201327). Image novelty detection based on mean-shift and typical set size. Proceedings of the 21th International Conference on Image Analysis and Processing, Lecce, Italy.","DOI":"10.1007\/978-3-031-06430-2_63"},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Hotelling, H. (1992). The generalization of Student\u2019s ratio. Breakthroughs in Statistics, Springer.","DOI":"10.1007\/978-1-4612-0919-5_4"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"110","DOI":"10.3905\/jpm.2004.110","article-title":"Honey, I shrunk the sample covariance matrix","volume":"30","author":"Ledoit","year":"2004","journal-title":"J. Portf. Manag."},{"key":"ref_5","unstructured":"Tan, M., and Le, Q. (2019, January 9\u201315). Efficientnet: Rethinking model scaling for convolutional neural networks. Proceedings of the International Conference on Machine Learning, Long Beach, CA, USA."},{"key":"ref_6","unstructured":"Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., and Gelly, S. (2020). An image is worth 16x16 words: Transformers for image recognition at scale. arXiv."},{"key":"ref_7","doi-asserted-by":"crossref","unstructured":"Bergmann, P., Fauser, M., Sattlegger, D., and Steger, C. (2019, January 15\u201320). MVTec AD - A comprehensive real-world dataset for unsupervised anomaly detection. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, CA, USA.","DOI":"10.1109\/CVPR.2019.00982"},{"key":"ref_8","doi-asserted-by":"crossref","unstructured":"Tian, Y., Wang, Y., Krishnan, D., Tenenbaum, J.B., and Isola, P. (2020, January 23\u201328). Rethinking few-shot image classification: A good embedding is all you need?. Proceedings of the European Conference on Computer Vision, Glasgow, UK.","DOI":"10.1007\/978-3-030-58568-6_16"},{"key":"ref_9","first-page":"1","article-title":"Generalizing from a few examples: A survey on few-shot learning","volume":"53","author":"Wang","year":"2020","journal-title":"ACM Comput. Surv. (csur)"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Pang, G., Cao, L., Chen, L., and Liu, H. (2018, January 19\u201323). Learning representations of ultrahigh-dimensional data for random distance-based outlier detection. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, London, UK.","DOI":"10.1145\/3219819.3220042"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"215","DOI":"10.1016\/j.sigpro.2013.12.026","article-title":"A review of novelty detection","volume":"99","author":"Pimentel","year":"2014","journal-title":"Signal Process."},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Rippel, O., Mertens, P., and Merhof, D. (2021, January 10\u201315). Modeling the distribution of normal data in pre-trained deep features for anomaly detection. Proceedings of the 2020 25th International Conference on Pattern Recognition (ICPR), Milan, Italy.","DOI":"10.1109\/ICPR48806.2021.9412109"},{"key":"ref_13","unstructured":"Ruff, L., Vandermeulen, R., Goernitz, N., Deecke, L., Siddiqui, S.A., Binder, A., M\u00fcller, E., and Kloft, M. (2018, January 10\u201315). Deep one-class classification. Proceedings of the International Conference on Machine Learning, Stockholm, Sweden."},{"key":"ref_14","unstructured":"Goyal, S., Raghunathan, A., Jain, M., Simhadri, H.V., and Jain, P. (2020, January 13\u201318). DROCC: Deep robust one-class classification. Proceedings of the International Conference on Machine Learning, Virtual."},{"key":"ref_15","first-page":"1","article-title":"Variational autoencoder based anomaly detection using reconstruction probability","volume":"2","author":"An","year":"2015","journal-title":"Spec. Lect. IE"},{"key":"ref_16","doi-asserted-by":"crossref","unstructured":"Rudolph, M., Wandt, B., and Rosenhahn, B. (2021, January 3\u20138). Same same but differnet: Semi-supervised defect detection with normalizing flows. Proceedings of the IEEE\/CVF Winter Conference on Applications of Computer Vision, Waikoloa, HI, USA.","DOI":"10.1109\/WACV48630.2021.00195"},{"key":"ref_17","unstructured":"Zhang, L., Goldstein, M., and Ranganath, R. (2021, January 18\u201324). Understanding failures in out-of-distribution detection with deep generative models. Proceedings of the International Conference on Machine Learning, Virtual."},{"key":"ref_18","unstructured":"Hyvarinen, A., and Morioka, H. (2016). Unsupervised feature extraction by time-contrastive learning and nonlinear ica. Adv. Neural Inf. Process. Syst., 29."},{"key":"ref_19","unstructured":"Hyvarinen, A., Sasaki, H., and Turner, R. (2019, January 16\u201318). Nonlinear ICA using auxiliary variables and generalized contrastive learning. Proceedings of the The 22nd International Conference on Artificial Intelligence and Statistics."},{"key":"ref_20","unstructured":"Golan, I., and El-Yaniv, R. (2018). Deep anomaly detection using geometric transformations. Adv. Neural Inf. Process. Syst., 31."},{"key":"ref_21","unstructured":"Gidaris, S., Singh, P., and Komodakis, N. (2018). Unsupervised representation learning by predicting image rotations. arXiv."},{"key":"ref_22","doi-asserted-by":"crossref","unstructured":"Roth, K., Pemula, L., Zepeda, J., Sch\u00f6lkopf, B., Brox, T., and Gehler, P. (2022, January 18\u201324). Towards total recall in industrial anomaly detection. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, USA.","DOI":"10.1109\/CVPR52688.2022.01392"},{"key":"ref_23","unstructured":"Sener, O., and Savarese, S. (2017). Active learning for convolutional neural networks: A core-set approach. arXiv."},{"key":"ref_24","doi-asserted-by":"crossref","unstructured":"Li, C.L., Sohn, K., Yoon, J., and Pfister, T. (2021, January 20\u201325). Cutpaste: Self-supervised learning for anomaly detection and localization. Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition, Nashville, TN, USA.","DOI":"10.1109\/CVPR46437.2021.00954"},{"key":"ref_25","unstructured":"Yi, J., and Yoon, S. (December, January 30). Patch svdd: Patch-level svdd for anomaly detection and segmentation. Proceedings of the Asian Conference on Computer Vision, Kyoto, Japan."},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Defard, T., Setkov, A., Loesch, A., and Audigier, R. (2021, January 10\u201315). Padim: A patch distribution modeling framework for anomaly detection and localization. Proceedings of the International Conference on Pattern Recognition, Virtual.","DOI":"10.1007\/978-3-030-68799-1_35"},{"key":"ref_27","first-page":"49","article-title":"On the generalised distance in statistics","volume":"2","author":"Mahalanobis","year":"1936","journal-title":"Proc. Natl. Inst. Sci. India"},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Zagoruyko, S., and Komodakis, N. (2016). Wide residual networks. arXiv.","DOI":"10.5244\/C.30.87"},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"211","DOI":"10.1007\/s11263-015-0816-y","article-title":"Imagenet large scale visual recognition challenge","volume":"115","author":"Russakovsky","year":"2015","journal-title":"Int. J. Comput. Vis."},{"key":"ref_30","unstructured":"Luo, W., Li, Y., Urtasun, R., and Zemel, R. (2016). Understanding the effective receptive field in deep convolutional neural networks. Adv. Neural Inf. Process. Syst., 29."},{"key":"ref_31","unstructured":"Sch\u00f6lkopf, B., Williamson, R.C., Smola, A., Shawe-Taylor, J., and Platt, J. (1999). Support vector method for novelty detection. Adv. Neural Inf. Process. Syst., 12."},{"key":"ref_32","unstructured":"Krizhevsky, A., and Hinton, G. (2009). Learning Multiple Layers of Features from Tiny Images, Citeseer."},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"e21","DOI":"10.23915\/distill.00021","article-title":"Computing Receptive Fields of Convolutional Neural Networks","volume":"4","author":"Araujo","year":"2019","journal-title":"Distill"},{"key":"ref_34","doi-asserted-by":"crossref","unstructured":"Sheynin, S., Benaim, S., and Wolf, L. (2021, January 11\u201317). A hierarchical transformation-discriminating generative model for few shot anomaly detection. Proceedings of the IEEE\/CVF International Conference on Computer Vision, Montreal, QC, Canada.","DOI":"10.1109\/ICCV48922.2021.00838"},{"key":"ref_35","doi-asserted-by":"crossref","first-page":"287","DOI":"10.1016\/0165-1684(94)90029-9","article-title":"Independent component analysis, a new concept?","volume":"36","author":"Comon","year":"1994","journal-title":"Signal Process."}],"container-title":["Sensors"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/1424-8220\/22\/19\/7674\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,10,11]],"date-time":"2025-10-11T00:49:06Z","timestamp":1760143746000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/1424-8220\/22\/19\/7674"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2022,10,10]]},"references-count":35,"journal-issue":{"issue":"19","published-online":{"date-parts":[[2022,10]]}},"alternative-id":["s22197674"],"URL":"https:\/\/doi.org\/10.3390\/s22197674","relation":{},"ISSN":["1424-8220"],"issn-type":[{"value":"1424-8220","type":"electronic"}],"subject":[],"published":{"date-parts":[[2022,10,10]]}}}