{"url":"/sota/unsupervised-semantic-segmentation-with-4","task":{"name":"Unsupervised Semantic Segmentation with Language-image Pre-training","url":"/task/unsupervised-semantic-segmentation-with","note":null},"dataset":{"name":"ADE20K","url":"/dataset/ade20k"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"A segmentation task which does not utilise any human-level supervision for semantic segmentation except for a backbone which is initialised with features pre-trained with image-level labels.","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":["Mean IoU (val)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Mean IoU (val)":"higher"}},"counts":{"rows":13,"rows_with_code":13,"rows_with_paper_page":13,"rows_dated":13,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"CorrCLIP","metrics":{"Mean IoU (val)":"30.7"},"uses_additional_data":false,"paper_date":"2024-11-15","paper":"/paper/corrclip-reconstructing-correlations-in-clip","paper_url":"https://arxiv.org/abs/2411.10086v1","paper_title":"CorrCLIP: Reconstructing Correlations in CLIP with Off-the-Shelf Foundation Models for Open-Vocabulary Semantic Segmentation","code":"https://github.com/zdk258/CorrCLIP","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":9,"n_samples":9,"n_pointer_only_licence":9}},{"rank_in_archive_order":2,"model":"TextRegion","metrics":{"Mean IoU (val)":"27.3"},"uses_additional_data":false,"paper_date":"2025-05-29","paper":"/paper/textregion-text-aligned-region-tokens-from","paper_url":"https://arxiv.org/abs/2505.23769v1","paper_title":"TextRegion: Text-Aligned Region Tokens from Frozen Image-Text Models","code":"https://github.com/avaxiao/TextRegion","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"Trident","metrics":{"Mean IoU (val)":"26.7"},"uses_additional_data":false,"paper_date":"2024-11-14","paper":"/paper/harnessing-vision-foundation-models-for-high","paper_url":"https://arxiv.org/abs/2411.09219v1","paper_title":"Harnessing Vision Foundation Models for High-Performance, Training-Free Open Vocabulary Segmentation","code":"https://github.com/YuHengsss/Trident","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"ProxyCLIP","metrics":{"Mean IoU (val)":"24.2"},"uses_additional_data":false,"paper_date":"2024-08-09","paper":"/paper/proxyclip-proxy-attention-improves-clip-for","paper_url":"https://arxiv.org/abs/2408.04883v1","paper_title":"ProxyCLIP: Proxy Attention Improves CLIP for Open-Vocabulary Segmentation","code":"https://github.com/mc-lan/proxyclip","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":6,"n_samples":9,"n_pointer_only_licence":9}},{"rank_in_archive_order":5,"model":"COSMOS ViT-B/16","metrics":{"Mean IoU (val)":"17.7"},"uses_additional_data":false,"paper_date":"2024-12-02","paper":"/paper/cosmos-cross-modality-self-distillation-for","paper_url":"https://arxiv.org/abs/2412.01814v2","paper_title":"COSMOS: Cross-Modality Self-Distillation for Vision Language Pre-training","code":"https://github.com/ExplainableML/cosmos","n_code_links":1,"syntology":null},{"rank_in_archive_order":6,"model":"TagAlign","metrics":{"Mean IoU (val)":"17.3"},"uses_additional_data":false,"paper_date":"2023-12-21","paper":"/paper/tagalign-improving-vision-language-alignment","paper_url":"https://arxiv.org/abs/2312.14149v4","paper_title":"TagAlign: Improving Vision-Language Alignment with Multi-Tag Classification","code":"https://github.com/Qinying-Liu/TagAlign","n_code_links":1,"syntology":null},{"rank_in_archive_order":7,"model":"TCL","metrics":{"Mean IoU (val)":"17.1"},"uses_additional_data":false,"paper_date":"2022-12-01","paper":"/paper/learning-to-generate-text-grounded-mask-for","paper_url":"https://arxiv.org/abs/2212.00785v2","paper_title":"Learning to Generate Text-grounded Mask for Open-world Semantic Segmentation from Only Image-Text Pairs","code":"https://github.com/kakaobrain/tcl","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":7,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":8,"model":"TTD (TCL)","metrics":{"Mean IoU (val)":"17.0"},"uses_additional_data":false,"paper_date":"2024-03-30","paper":"/paper/ttd-text-tag-self-distillation-enhancing","paper_url":"https://arxiv.org/abs/2404.00384v2","paper_title":"TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias","code":"https://github.com/shjo-april/TTD","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":9,"model":"CLIPpy ViT-B","metrics":{"Mean IoU (val)":"13.5"},"uses_additional_data":false,"paper_date":"2022-10-18","paper":"/paper/perceptual-grouping-in-vision-language-models","paper_url":"https://arxiv.org/abs/2210.09996v3","paper_title":"Perceptual Grouping in Contrastive Vision-Language Models","code":"https://github.com/kahnchana/clippy","n_code_links":2,"syntology":null},{"rank_in_archive_order":10,"model":"TTD (MaskCLIP)","metrics":{"Mean IoU (val)":"12.7"},"uses_additional_data":false,"paper_date":"2024-03-30","paper":"/paper/ttd-text-tag-self-distillation-enhancing","paper_url":"https://arxiv.org/abs/2404.00384v2","paper_title":"TTD: Text-Tag Self-Distillation Enhancing Image-Text Alignment in CLIP to Alleviate Single Tag Bias","code":"https://github.com/shjo-april/TTD","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}},{"rank_in_archive_order":11,"model":"ReCo","metrics":{"Mean IoU (val)":"11.2"},"uses_additional_data":false,"paper_date":"2022-06-14","paper":"/paper/reco-retrieve-and-co-segment-for-zero-shot-1","paper_url":"https://arxiv.org/abs/2206.07045v1","paper_title":"ReCo: Retrieve and Co-segment for Zero-shot Transfer","code":"https://github.com/NoelShin/reco","n_code_links":2,"syntology":null},{"rank_in_archive_order":12,"model":"MaskCLIP","metrics":{"Mean IoU (val)":"9.8"},"uses_additional_data":false,"paper_date":"2021-12-02","paper":"/paper/denseclip-extract-free-dense-labels-from-clip","paper_url":"https://arxiv.org/abs/2112.01071v2","paper_title":"Extract Free Dense Labels from CLIP","code":"https://github.com/chongzhou96/maskclip","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"GroupViT (RedCaps)","metrics":{"Mean IoU (val)":"9.2"},"uses_additional_data":false,"paper_date":"2022-02-22","paper":"/paper/groupvit-semantic-segmentation-emerges-from","paper_url":"https://arxiv.org/abs/2202.11094v5","paper_title":"GroupViT: Semantic Segmentation Emerges from Text Supervision","code":"https://github.com/huggingface/transformers","n_code_links":6,"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":5,"rows_with_any_sample_ran":4,"distinct_papers_with_graph_line":4,"distinct_papers_with_any_sample_ran":3,"samples_over_distinct_papers":{"n_ran":5,"n_unverified":22,"n_samples":27,"n_pointer_only_licence":19,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":6,"n_unverified":22,"n_samples":28,"n_pointer_only_licence":20,"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"}}}