{"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/consistency-regularization-and-cutmix-for","title":"Semi-supervised semantic segmentation needs strong, varied perturbations","arxiv_id":"1906.01916","date":"2019-06-05","proceeding":null,"authors":["Geoff French","Samuli Laine","Timo Aila","Michal Mackiewicz","Graham Finlayson"],"abstract":"Consistency regularization describes a class of approaches that have yielded ground breaking results in semi-supervised classification problems. Prior work has established the cluster assumption - under which the data distribution consists of uniform class clusters of samples separated by low density regions - as important to its success. We analyze the problem of semantic segmentation and find that its' distribution does not exhibit low density regions separating classes and offer this as an explanation for why semi-supervised segmentation is a challenging problem, with only a few reports of success. We then identify choice of augmentation as key to obtaining reliable performance without such low-density regions. We find that adapted variants of the recently proposed CutOut and CutMix augmentation techniques yield state-of-the-art semi-supervised semantic segmentation results in standard datasets. Furthermore, given its challenging nature we propose that semantic segmentation acts as an effective acid test for evaluating semi-supervised regularizers. Implementation at: https://github.com/Britefury/cutmix-semisup-seg.","url_abs":"https://arxiv.org/abs/1906.01916v5","url_pdf":"https://arxiv.org/pdf/1906.01916v5.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":"consistency-regularization-and-cutmix-for","repo_url":"https://github.com/Britefury/cutmix-semisup-seg","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"consistency-regularization-and-cutmix-for","repo_url":"https://github.com/ZHKKKe/PixelSSL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"consistency-regularization-and-cutmix-for","repo_url":"https://github.com/lorenmt/reco","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"consistency-regularization-and-cutmix-for","repo_url":"https://github.com/moucheng2017/learning_morphological_perturbation_ssl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"consistency-regularization-and-cutmix-for","repo_url":"https://github.com/moucheng2017/mismatchssl","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"semi-supervised-semantic-segmentation","task_name":"Semi-Supervised Semantic Segmentation"}],"methods":[{"method_slug":"cutmix","method_name":"CutMix"},{"method_slug":"cutout","method_name":"Cutout"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-42","task":"Semi-Supervised Semantic Segmentation","dataset":"ADE20K 1/16 labeled","model":"CutMix","rank_in_archive_order":5,"of":5,"metrics":{"Validation mIoU":"29.8"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-41","task":"Semi-Supervised Semantic Segmentation","dataset":"ADE20K 1/32 labeled","model":"CutMix","rank_in_archive_order":5,"of":5,"metrics":{"Validation mIoU":"26.2"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-3","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 100 samples labeled","model":"CutMix (DeepLab v2, ImageNet pre-trained)","rank_in_archive_order":13,"of":13,"metrics":{"Validation mIoU":"51.2"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-2","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 12.5% labeled","model":"CutMix (DeepLab v2, ImageNet pre-trained)","rank_in_archive_order":31,"of":33,"metrics":{"Validation mIoU":"60.34%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-1","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 25% labeled","model":"CutMix (DeepLab v2, ImageNet pre-trained)","rank_in_archive_order":27,"of":30,"metrics":{"Validation mIoU":"63.87%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-7","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 1% labeled","model":"CutMix (DeepLab v3+ ImageNet pre-trained)","rank_in_archive_order":4,"of":6,"metrics":{"Validation mIoU":"59.52%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-7","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 1% labeled","model":"CutMix (DeepLab v2 ImageNet pre-trained)","rank_in_archive_order":6,"of":6,"metrics":{"Validation mIoU":"53.79%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-4","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 12.5% labeled","model":"CutMix","rank_in_archive_order":27,"of":38,"metrics":{"Validation mIoU":"72.45%"},"uses_additional_data":true},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-4","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 12.5% labeled","model":"CutMix","rank_in_archive_order":34,"of":38,"metrics":{"Validation mIoU":"67.6%"},"uses_additional_data":true},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-6","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 2% labeled","model":"CutMix (DeepLab v3+ ImageNet pre-trained)","rank_in_archive_order":5,"of":12,"metrics":{"Validation mIoU":"67.05%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-6","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 2% labeled","model":"CutMix (DeepLab v2 ImageNet pre-trained)","rank_in_archive_order":8,"of":12,"metrics":{"Validation mIoU":"64.81%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-5","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 5% labeled","model":"CutMix (DeepLab v3+ ImageNet pre-trained)","rank_in_archive_order":6,"of":14,"metrics":{"Validation mIoU":"69.57%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-5","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 5% labeled","model":"CutMix (DeepLab v2 ImageNet pre-trained)","rank_in_archive_order":12,"of":14,"metrics":{"Validation mIoU":"66.48%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-23","task":"Semi-Supervised Semantic Segmentation","dataset":"ScribbleKITTI","model":"CutMix-Seg (Range View)","rank_in_archive_order":5,"of":9,"metrics":{"mIoU (1% Labels)":"36.7","mIoU (10% Labels)":"50.7","mIoU (20% Labels)":"52.9","mIoU (50% Labels)":"54.3"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-24","task":"Semi-Supervised Semantic Segmentation","dataset":"SemanticKITTI","model":"CutMix-Seg (Range View)","rank_in_archive_order":11,"of":12,"metrics":{"mIoU (1% Labels)":"37.4","mIoU (10% Labels)":"54.3","mIoU (20% Labels)":"56.6","mIoU (50% Labels)":"57.6"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-25","task":"Semi-Supervised Semantic Segmentation","dataset":"nuScenes","model":"CutMix-Seg (Range View)","rank_in_archive_order":7,"of":11,"metrics":{"mIoU (1% Labels)":"43.8","mIoU (10% Labels)":"63.9","mIoU (20% Labels)":"64.8","mIoU (50% Labels)":"69.8"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1906.01916","atlas_url":"https://app.syntology.ai/?focus=1906.01916","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"1906.01916"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-25T09:33:49+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/lorenmt/reco","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/Britefury/cutmix-semisup-seg","reach":{"status":"ok","spdx":"MIT"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/moucheng2017/learning_morphological_perturbation_ssl","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/moucheng2017/mismatchssl","reach":{"status":"ok"}},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/ZHKKKe/PixelSSL","reach":null}],"summary":{"unverified":3},"by_repo_kind":{"official":{"samples":3,"ran":0,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"06336a1a4a1474b9","entry":"convert","repo":"Britefury/cutmix-semisup-seg","repo_kind":"official","path":"convert_cityscapes.py","file_url":"https://github.com/Britefury/cutmix-semisup-seg/blob/HEAD/convert_cityscapes.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"06336a1a4a1474b9"}},{"code_sha256_prefix":"70a30737ad42c413","entry":"convert_isic","repo":"Britefury/cutmix-semisup-seg","repo_kind":"official","path":"convert_isic.py","file_url":"https://github.com/Britefury/cutmix-semisup-seg/blob/HEAD/convert_isic.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"70a30737ad42c413"}},{"code_sha256_prefix":"07cb1ccb51e03d35","entry":"get_mean_std","repo":"Britefury/cutmix-semisup-seg","repo_kind":"official","path":"datapipe/seg_transforms.py","file_url":"https://github.com/Britefury/cutmix-semisup-seg/blob/HEAD/datapipe/seg_transforms.py","link_basis":"first_harvest_node","language":"python","status":"unverified","verification_level":0,"contract_check":null,"metamorphic_tier":null,"behaviour_fingerprint":false,"licence":"MIT","inline_ok":true,"mcp_get_code":{"code_sha256":"07cb1ccb51e03d35"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}