{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/far-away-in-the-deep-space-nearest-neighbor","title":"Far Away in the Deep Space: Dense Nearest-Neighbor-Based Out-of-Distribution Detection","arxiv_id":"2211.06660","date":"2022-11-12","proceeding":null,"authors":["Silvio Galesso","Max Argus","Thomas Brox"],"abstract":"The key to out-of-distribution detection is density estimation of the in-distribution data or of its feature representations. This is particularly challenging for dense anomaly detection in domains where the in-distribution data has a complex underlying structure. Nearest-Neighbors approaches have been shown to work well in object-centric data domains, such as industrial inspection and image classification. In this paper, we show that nearest-neighbor approaches also yield state-of-the-art results on dense novelty detection in complex driving scenes when working with an appropriate feature representation. In particular, we find that transformer-based architectures produce representations that yield much better similarity metrics for the task. We identify the multi-head structure of these models as one of the reasons, and demonstrate a way to transfer some of the improvements to CNNs. Ultimately, the approach is simple and non-invasive, i.e., it does not affect the primary segmentation performance, refrains from training on examples of anomalies, and achieves state-of-the-art results on RoadAnomaly, StreetHazards, and SegmentMeIfYouCan-Anomaly.","url_abs":"https://arxiv.org/abs/2211.06660v2","url_pdf":"https://arxiv.org/pdf/2211.06660v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"far-away-in-the-deep-space-nearest-neighbor","repo_url":"https://github.com/silviogalesso/dense-ood-knns","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"anomaly-detection","task_name":"Anomaly Detection"},{"task_slug":"density-estimation","task_name":"Density Estimation"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"novelty-detection","task_name":"Novelty Detection"},{"task_slug":"out-of-distribution-detection","task_name":"Out-of-Distribution Detection"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/anomaly-detection-on-fishyscapes-l-f","task":"Anomaly Detection","dataset":"Fishyscapes L&F","model":"cDNP+OE","rank_in_archive_order":1,"of":18,"metrics":{"AP":"69.8","FPR95":"7.5"},"uses_additional_data":true},{"leaderboard":"/sota/anomaly-detection-on-fishyscapes-l-f","task":"Anomaly Detection","dataset":"Fishyscapes L&F","model":"cDNP","rank_in_archive_order":3,"of":18,"metrics":{"AP":"62.2","FPR95":"8.9"},"uses_additional_data":false},{"leaderboard":"/sota/anomaly-detection-on-road-anomaly","task":"Anomaly Detection","dataset":"Road Anomaly","model":"cDNP","rank_in_archive_order":4,"of":10,"metrics":{"AP":"85.6","FPR95":"9.8"},"uses_additional_data":false},{"leaderboard":"/sota/out-of-distribution-detection-on-ade-ood","task":"Out-of-Distribution Detection","dataset":"ADE-OoD","model":"cDNP","rank_in_archive_order":3,"of":4,"metrics":{"AP":"62.35","FPR@95":"39.20"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}