{"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/dense-fixmatch-a-simple-semi-supervised","title":"Dense FixMatch: a simple semi-supervised learning method for pixel-wise prediction tasks","arxiv_id":"2210.09919","date":"2022-10-18","proceeding":null,"authors":["Miquel Martí i Rabadán","Alessandro Pieropan","Hossein Azizpour","Atsuto Maki"],"abstract":"We propose Dense FixMatch, a simple method for online semi-supervised learning of dense and structured prediction tasks combining pseudo-labeling and consistency regularization via strong data augmentation. We enable the application of FixMatch in semi-supervised learning problems beyond image classification by adding a matching operation on the pseudo-labels. This allows us to still use the full strength of data augmentation pipelines, including geometric transformations. We evaluate it on semi-supervised semantic segmentation on Cityscapes and Pascal VOC with different percentages of labeled data and ablate design choices and hyper-parameters. Dense FixMatch significantly improves results compared to supervised learning using only labeled data, approaching its performance with 1/4 of the labeled samples.","url_abs":"https://arxiv.org/abs/2210.09919v1","url_pdf":"https://arxiv.org/pdf/2210.09919v1.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":"dense-fixmatch-a-simple-semi-supervised","repo_url":"https://github.com/miquelmarti/DenseFixMatch","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"data-augmentation","task_name":"Data Augmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"semi-supervised-semantic-segmentation","task_name":"Semi-Supervised Semantic Segmentation"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"fixmatch","method_name":"FixMatch"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-2","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 12.5% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-101, uniform sampling, single pass eval)","rank_in_archive_order":20,"of":33,"metrics":{"Validation mIoU":"73.91%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-2","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 12.5% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-50, uniform sampling, single pass eval)","rank_in_archive_order":21,"of":33,"metrics":{"Validation mIoU":"73.39%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-22","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 6.25% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":15,"of":18,"metrics":{"Validation mIoU":"71.1%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-22","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 6.25% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-50, uniform sampling, single pass eval)","rank_in_archive_order":16,"of":18,"metrics":{"Validation mIoU":"70.65%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-35","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 93 labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":2,"of":3,"metrics":{"Validation mIoU":"66.97"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-35","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes 93 labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single-pass eval)","rank_in_archive_order":3,"of":3,"metrics":{"Validation mIoU":"65.81"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-40","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes with extra (no coarse labels)","model":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":1,"of":2,"metrics":{"Validation mIoU":"80.82"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-40","task":"Semi-Supervised Semantic Segmentation","dataset":"Cityscapes with extra (no coarse labels)","model":"Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single pass eval)","rank_in_archive_order":2,"of":2,"metrics":{"Validation mIoU":"79.98"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-9","task":"Semi-Supervised Semantic Segmentation","dataset":"PASCAL VOC 2012 25% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":26,"of":27,"metrics":{"Validation mIoU":"72.04"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-9","task":"Semi-Supervised Semantic Segmentation","dataset":"PASCAL VOC 2012 25% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single pass eval)","rank_in_archive_order":27,"of":27,"metrics":{"Validation mIoU":"69.02"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-15","task":"Semi-Supervised Semantic Segmentation","dataset":"PASCAL VOC 2012 50%","model":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":13,"of":14,"metrics":{"Validation mIoU":"74.73%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-15","task":"Semi-Supervised Semantic Segmentation","dataset":"PASCAL VOC 2012 50%","model":"Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single pass eval)","rank_in_archive_order":14,"of":14,"metrics":{"Validation mIoU":"71.69%"},"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":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":36,"of":38,"metrics":{"Validation mIoU":"65.82%"},"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":"Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single pass eval)","rank_in_archive_order":38,"of":38,"metrics":{"Validation mIoU":"62.49%"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-21","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 6.25% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-101, over-sampling, single pass eval)","rank_in_archive_order":18,"of":19,"metrics":{"Validation mIoU":"54.85"},"uses_additional_data":false},{"leaderboard":"/sota/semi-supervised-semantic-segmentation-on-21","task":"Semi-Supervised Semantic Segmentation","dataset":"Pascal VOC 2012 6.25% labeled","model":"Dense FixMatch (DeepLabv3+ ResNet-50, over-sampling, single pass eval)","rank_in_archive_order":19,"of":19,"metrics":{"Validation mIoU":"52.15"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}