{"url":"/sota/unsupervised-semantic-segmentation-on","task":{"name":"Unsupervised Semantic Segmentation","url":"/task/unsupervised-semantic-segmentation","note":null},"dataset":{"name":"Cityscapes test","url":"/dataset/cityscapes"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"Models that learn to segment each image (i.e. assign a class to every pixel) without seeing the ground truth labels.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [SegSort: Segmentation by Discriminative Sorting of Segments](http://openaccess.thecvf.com/content_ICCV_2019/papers/Hwang_SegSort_Segmentation_by_Discriminative_Sorting_of_Segments_ICCV_2019_paper.pdf) )</span>","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["mIoU","Accuracy","Pixel Accuracy"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"mIoU":null,"Accuracy":"higher","Pixel Accuracy":"higher"}},"counts":{"rows":14,"rows_with_code":13,"rows_with_paper_page":14,"rows_dated":14,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"CUPS","metrics":{"Accuracy":"83.2 ","mIoU":"26.8"},"uses_additional_data":false,"paper_date":"2025-04-02","paper":"/paper/scene-centric-unsupervised-panoptic","paper_url":"https://arxiv.org/abs/2504.01955v1","paper_title":"Scene-Centric Unsupervised Panoptic Segmentation","code":"https://github.com/visinf/cups","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":8,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"ViCE","metrics":{"Accuracy":"84.3","mIoU":"25.2"},"uses_additional_data":false,"paper_date":"2021-11-24","paper":"/paper/vice-self-supervised-visual-concept","paper_url":"https://arxiv.org/abs/2111.12460v3","paper_title":"ViCE: Improving Dense Representation Learning by Superpixelization and Contrasting Cluster Assignment","code":"https://github.com/robin-karlsson0/vice","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"EAGLE (DINO, ViT-B/8)","metrics":{"Accuracy":"79.4","mIoU":"22.1"},"uses_additional_data":false,"paper_date":"2024-03-03","paper":"/paper/eagle-eigen-aggregation-learning-for-object","paper_url":"https://arxiv.org/abs/2403.01482v4","paper_title":"EAGLE: Eigen Aggregation Learning for Object-Centric Unsupervised Semantic Segmentation","code":"https://github.com/MICV-yonsei/EAGLE","n_code_links":1,"syntology":{"n_ran":14,"n_unverified":5,"n_samples":19,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"EQUSS","metrics":{"Accuracy":"79.9","mIoU":"22.0"},"uses_additional_data":false,"paper_date":"2023-12-12","paper":"/paper/expand-and-quantize-unsupervised-semantic","paper_url":"https://arxiv.org/abs/2312.07342v1","paper_title":"Expand-and-Quantize: Unsupervised Semantic Segmentation Using High-Dimensional Space and Product Quantization","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"PriMaPs-EM + STEGO (DINO ViT-B/8)","metrics":{"Accuracy":"78.6","mIoU":"21.6"},"uses_additional_data":false,"paper_date":"2024-04-25","paper":"/paper/boosting-unsupervised-semantic-segmentation","paper_url":"https://arxiv.org/abs/2404.16818v2","paper_title":"Boosting Unsupervised Semantic Segmentation with Principal Mask Proposals","code":"https://github.com/visinf/primaps","n_code_links":1,"syntology":null},{"rank_in_archive_order":6,"model":"STEGO","metrics":{"Accuracy":"73.2","mIoU":"21.0"},"uses_additional_data":false,"paper_date":"2022-03-16","paper":"/paper/unsupervised-semantic-segmentation-by-2","paper_url":"https://arxiv.org/abs/2203.08414v1","paper_title":"Unsupervised Semantic Segmentation by Distilling Feature Correspondences","code":"https://github.com/mhamilton723/STEGO","n_code_links":3,"syntology":{"n_ran":8,"n_unverified":6,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"EAGLE (DINO, ViT-S/8)","metrics":{"Accuracy":"81.8","mIoU":"19.7"},"uses_additional_data":false,"paper_date":"2024-03-03","paper":"/paper/eagle-eigen-aggregation-learning-for-object","paper_url":"https://arxiv.org/abs/2403.01482v4","paper_title":"EAGLE: Eigen Aggregation Learning for Object-Centric Unsupervised Semantic Segmentation","code":"https://github.com/MICV-yonsei/EAGLE","n_code_links":1,"syntology":{"n_ran":14,"n_unverified":5,"n_samples":19,"n_pointer_only_licence":0}},{"rank_in_archive_order":8,"model":"PriMaPs-EM (DINO ViT-S/8)","metrics":{"Accuracy":"81.2","mIoU":"19.4"},"uses_additional_data":false,"paper_date":"2024-04-25","paper":"/paper/boosting-unsupervised-semantic-segmentation","paper_url":"https://arxiv.org/abs/2404.16818v2","paper_title":"Boosting Unsupervised Semantic Segmentation with Principal Mask Proposals","code":"https://github.com/visinf/primaps","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"HP","metrics":{"Accuracy":"80.1","mIoU":"18.4"},"uses_additional_data":false,"paper_date":"2023-03-27","paper":"/paper/leveraging-hidden-positives-for-unsupervised","paper_url":"https://arxiv.org/abs/2303.15014v1","paper_title":"Leveraging Hidden Positives for Unsupervised Semantic Segmentation","code":"https://github.com/hynnsk/hp","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":2,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":10,"model":"GraPix + AUT","metrics":{"mIoU":"14.54"},"uses_additional_data":false,"paper_date":"2024-12-04","paper":"/paper/grapix-exploring-graph-modularity","paper_url":"https://doi.org/10.1007/978-3-031-78192-6_13","paper_title":"GraPix: Exploring Graph Modularity Optimization for Unsupervised Pixel Clustering","code":"https://github.com/SonalKumar95/GraPix","n_code_links":1,"syntology":null},{"rank_in_archive_order":11,"model":"Grapix","metrics":{"mIoU":"14.33"},"uses_additional_data":false,"paper_date":"2024-12-04","paper":"/paper/grapix-exploring-graph-modularity","paper_url":"https://doi.org/10.1007/978-3-031-78192-6_13","paper_title":"GraPix: Exploring Graph Modularity Optimization for Unsupervised Pixel Clustering","code":"https://github.com/SonalKumar95/GraPix","n_code_links":1,"syntology":null},{"rank_in_archive_order":12,"model":"PiCIE","metrics":{"Accuracy":"65.5","mIoU":"12.3"},"uses_additional_data":false,"paper_date":"2021-03-30","paper":"/paper/picie-unsupervised-semantic-segmentation","paper_url":"https://arxiv.org/abs/2103.17070v1","paper_title":"PiCIE: Unsupervised Semantic Segmentation using Invariance and Equivariance in Clustering","code":"https://github.com/janghyuncho/PiCIE","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":2,"n_samples":4,"n_pointer_only_licence":1}},{"rank_in_archive_order":13,"model":"MDC","metrics":{"Accuracy":"40.7","mIoU":"7.1"},"uses_additional_data":false,"paper_date":"2018-07-15","paper":"/paper/deep-clustering-for-unsupervised-learning-of","paper_url":"http://arxiv.org/abs/1807.05520v2","paper_title":"Deep Clustering for Unsupervised Learning of Visual Features","code":"https://github.com/facebookresearch/deepcluster","n_code_links":9,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":4}},{"rank_in_archive_order":14,"model":"GraPix","metrics":{"Pixel Accuracy":"64.89"},"uses_additional_data":false,"paper_date":"2024-12-04","paper":"/paper/grapix-exploring-graph-modularity","paper_url":"https://doi.org/10.1007/978-3-031-78192-6_13","paper_title":"GraPix: Exploring Graph Modularity Optimization for Unsupervised Pixel Clustering","code":"https://github.com/SonalKumar95/GraPix","n_code_links":1,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,264 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6264,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":3316},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":7,"rows_with_any_sample_ran":6,"distinct_papers_with_graph_line":6,"distinct_papers_with_any_sample_ran":5,"samples_over_distinct_papers":{"n_ran":30,"n_unverified":25,"n_samples":55,"n_pointer_only_licence":5,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":44,"n_unverified":30,"n_samples":74,"n_pointer_only_licence":5,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}