{"url":"/sota/semantic-segmentation-on-pascal-voc-2012-val","task":{"name":"Semantic Segmentation","url":"/task/semantic-segmentation","note":null},"dataset":{"name":"PASCAL VOC 2012 val","url":null},"category":"Computer Vision","categories":["Computer Code","Computer Vision","Medical","Robots"],"category_note":null,"description":null,"description_from":null,"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","Mean IoU","mIoU (Syn)"],"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,"Mean IoU":"higher","mIoU (Syn)":null}},"counts":{"rows":29,"rows_with_code":24,"rows_with_paper_page":29,"rows_dated":29,"rows_using_additional_data":2},"rows":[{"rank_in_archive_order":1,"model":"EfficientNet-L2+NAS-FPN (single scale test, with self-training)","metrics":{"mIoU":"90.0%"},"uses_additional_data":false,"paper_date":"2020-06-11","paper":"/paper/rethinking-pre-training-and-self-training","paper_url":"https://arxiv.org/abs/2006.06882v2","paper_title":"Rethinking Pre-training and Self-training","code":"https://github.com/tensorflow/tpu/tree/master/models/official/detection/projects/self_training","n_code_links":2,"syntology":null},{"rank_in_archive_order":2,"model":"TADP","metrics":{"mIoU":"87.11%"},"uses_additional_data":false,"paper_date":"2023-09-29","paper":"/paper/text-image-alignment-for-diffusion-based","paper_url":"https://arxiv.org/abs/2310.00031v3","paper_title":"Text-image Alignment for Diffusion-based Perception","code":"https://github.com/damaggu/tadp","n_code_links":2,"syntology":null},{"rank_in_archive_order":3,"model":"Eff-B7 NAS-FPN (Copy-Paste pre-training, single-scale))","metrics":{"mIoU":"86.6%"},"uses_additional_data":false,"paper_date":"2020-12-13","paper":"/paper/simple-copy-paste-is-a-strong-data","paper_url":"https://arxiv.org/abs/2012.07177v2","paper_title":"Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation","code":"https://github.com/PaddlePaddle/PaddleOCR","n_code_links":5,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"ExFuse (ResNeXt-131)","metrics":{"mIoU":"85.8%"},"uses_additional_data":true,"paper_date":"2018-04-11","paper":"/paper/exfuse-enhancing-feature-fusion-for-semantic","paper_url":"http://arxiv.org/abs/1804.03821v1","paper_title":"ExFuse: Enhancing Feature Fusion for Semantic Segmentation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"SpineNet-S143 (single-scale test)","metrics":{"mIoU":"85.64%"},"uses_additional_data":false,"paper_date":"2021-03-23","paper":"/paper/dilated-spinenet-for-semantic-segmentation","paper_url":"https://arxiv.org/abs/2103.12270v1","paper_title":"Dilated SpineNet for Semantic Segmentation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":6,"model":"DeepLabv3-JFT","metrics":{"mIoU":"82.7%"},"uses_additional_data":true,"paper_date":"2017-06-17","paper":"/paper/rethinking-atrous-convolution-for-semantic","paper_url":"http://arxiv.org/abs/1706.05587v3","paper_title":"Rethinking Atrous Convolution for Semantic Image Segmentation","code":"https://github.com/tensorflow/models","n_code_links":77,"syntology":{"n_ran":3,"n_unverified":4,"n_samples":7,"n_pointer_only_licence":3}},{"rank_in_archive_order":7,"model":"Auto-DeepLab-L","metrics":{"mIoU":"82.04%"},"uses_additional_data":false,"paper_date":"2019-01-10","paper":"/paper/auto-deeplab-hierarchical-neural-architecture","paper_url":"http://arxiv.org/abs/1901.02985v2","paper_title":"Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation","code":"https://github.com/tensorflow/models","n_code_links":12,"syntology":{"n_ran":1,"n_unverified":3,"n_samples":4,"n_pointer_only_licence":1}},{"rank_in_archive_order":8,"model":"ResNet-GCN","metrics":{"mIoU":"81.0%"},"uses_additional_data":false,"paper_date":"2017-03-08","paper":"/paper/large-kernel-matters-improve-semantic","paper_url":"http://arxiv.org/abs/1703.02719v1","paper_title":"Large Kernel Matters -- Improve Semantic Segmentation by Global Convolutional Network","code":"https://github.com/y-ouali/pytorch_segmentation","n_code_links":2,"syntology":null},{"rank_in_archive_order":9,"model":"HyperSeg-L","metrics":{"mIoU":"80.61%"},"uses_additional_data":false,"paper_date":"2020-12-21","paper":"/paper/hyperseg-patch-wise-hypernetwork-for-real","paper_url":"https://arxiv.org/abs/2012.11582v2","paper_title":"HyperSeg: Patch-wise Hypernetwork for Real-time Semantic Segmentation","code":"https://github.com/YuvalNirkin/hyperseg","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"DFN (ResNet-101)","metrics":{"mIoU":"80.60%"},"uses_additional_data":false,"paper_date":"2018-04-25","paper":"/paper/learning-a-discriminative-feature-network-for","paper_url":"http://arxiv.org/abs/1804.09337v1","paper_title":"Learning a Discriminative Feature Network for Semantic Segmentation","code":"https://github.com/ycszen/TorchSeg","n_code_links":3,"syntology":{"n_ran":0,"n_unverified":8,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"WASPnet-CRF (ours)","metrics":{"mIoU":"80.41%"},"uses_additional_data":false,"paper_date":"2019-12-06","paper":"/paper/waterfall-atrous-spatial-pooling-architecture","paper_url":"https://arxiv.org/abs/1912.03183v1","paper_title":"Waterfall Atrous Spatial Pooling Architecture for Efficient Semantic Segmentation","code":"https://github.com/bmartacho/WASP","n_code_links":1,"syntology":null},{"rank_in_archive_order":12,"model":"Deeplab v3+ (Res2Net-101)","metrics":{"mIoU":"79.3%"},"uses_additional_data":false,"paper_date":"2019-04-02","paper":"/paper/res2net-a-new-multi-scale-backbone","paper_url":"https://arxiv.org/abs/1904.01169v3","paper_title":"Res2Net: A New Multi-scale Backbone Architecture","code":"https://github.com/rwightman/pytorch-image-models","n_code_links":34,"syntology":{"n_ran":3,"n_unverified":6,"n_samples":9,"n_pointer_only_licence":9}},{"rank_in_archive_order":13,"model":"FastDenseNas-arch0","metrics":{"mIoU":"78.0%"},"uses_additional_data":false,"paper_date":"2018-10-25","paper":"/paper/fast-neural-architecture-search-of-compact","paper_url":"https://arxiv.org/abs/1810.10804v3","paper_title":"Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells","code":"https://github.com/mindspore-ai/models/tree/master/research/cv/adelaide_ea","n_code_links":4,"syntology":null},{"rank_in_archive_order":14,"model":"ReLICv2","metrics":{"mIoU":"77.9%"},"uses_additional_data":false,"paper_date":"2022-01-13","paper":"/paper/pushing-the-limits-of-self-supervised-resnets","paper_url":"https://arxiv.org/abs/2201.05119v2","paper_title":"Pushing the limits of self-supervised ResNets: Can we outperform supervised learning without labels on ImageNet?","code":"https://github.com/google-deepmind/relicv2","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":14,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":15,"model":"DeepLab-CRF (ResNet-101)","metrics":{"mIoU":"77.69%"},"uses_additional_data":false,"paper_date":"2016-06-02","paper":"/paper/deeplab-semantic-image-segmentation-with-deep","paper_url":"http://arxiv.org/abs/1606.00915v2","paper_title":"DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs","code":"https://github.com/tensorflow/models/tree/master/research/deeplab","n_code_links":47,"syntology":{"n_ran":28,"n_unverified":35,"n_samples":63,"n_pointer_only_licence":16}},{"rank_in_archive_order":16,"model":"FastDenseNas-arch2","metrics":{"mIoU":"77.3%"},"uses_additional_data":false,"paper_date":"2018-10-25","paper":"/paper/fast-neural-architecture-search-of-compact","paper_url":"https://arxiv.org/abs/1810.10804v3","paper_title":"Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells","code":"https://github.com/mindspore-ai/models/tree/master/research/cv/adelaide_ea","n_code_links":4,"syntology":null},{"rank_in_archive_order":17,"model":"DetCon","metrics":{"mIoU":"77.3%"},"uses_additional_data":false,"paper_date":"2022-01-13","paper":"/paper/pushing-the-limits-of-self-supervised-resnets","paper_url":"https://arxiv.org/abs/2201.05119v2","paper_title":"Pushing the limits of self-supervised ResNets: Can we outperform supervised learning without labels on ImageNet?","code":"https://github.com/google-deepmind/relicv2","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":14,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":18,"model":"FastDenseNas-arch1","metrics":{"mIoU":"77.1%"},"uses_additional_data":false,"paper_date":"2018-10-25","paper":"/paper/fast-neural-architecture-search-of-compact","paper_url":"https://arxiv.org/abs/1810.10804v3","paper_title":"Fast Neural Architecture Search of Compact Semantic Segmentation Models via Auxiliary Cells","code":"https://github.com/mindspore-ai/models/tree/master/research/cv/adelaide_ea","n_code_links":4,"syntology":null},{"rank_in_archive_order":19,"model":"DeepLabv3 (ImageNet+300M)","metrics":{"mIoU":"76.5%"},"uses_additional_data":false,"paper_date":"2017-07-10","paper":"/paper/revisiting-unreasonable-effectiveness-of-data","paper_url":"http://arxiv.org/abs/1707.02968v2","paper_title":"Revisiting Unreasonable Effectiveness of Data in Deep Learning Era","code":"https://github.com/Tencent/tencent-ml-images","n_code_links":2,"syntology":null},{"rank_in_archive_order":20,"model":"BYOL","metrics":{"mIoU":"75.7%"},"uses_additional_data":false,"paper_date":"2022-01-13","paper":"/paper/pushing-the-limits-of-self-supervised-resnets","paper_url":"https://arxiv.org/abs/2201.05119v2","paper_title":"Pushing the limits of self-supervised ResNets: Can we outperform supervised learning without labels on ImageNet?","code":"https://github.com/google-deepmind/relicv2","n_code_links":1,"syntology":{"n_ran":0,"n_unverified":14,"n_samples":14,"n_pointer_only_licence":0}},{"rank_in_archive_order":21,"model":"DiCENet","metrics":{"mIoU":"66.5%"},"uses_additional_data":false,"paper_date":"2019-06-08","paper":"/paper/dicenet-dimension-wise-convolutions-for","paper_url":"https://arxiv.org/abs/1906.03516v3","paper_title":"DiCENet: Dimension-wise Convolutions for Efficient Networks","code":"https://github.com/osmr/imgclsmob","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"RRM","metrics":{"mIoU":"66.3"},"uses_additional_data":false,"paper_date":"2019-11-19","paper":"/paper/reliability-does-matter-an-end-to-end-weakly","paper_url":"https://arxiv.org/abs/1911.08039v1","paper_title":"Reliability Does Matter: An End-to-End Weakly Supervised Semantic Segmentation Approach","code":"https://github.com/zbf1991/RRM","n_code_links":1,"syntology":null},{"rank_in_archive_order":23,"model":"SIW","metrics":{"mIoU":"65%"},"uses_additional_data":false,"paper_date":"2022-02-04","paper":"/paper/the-devil-is-in-the-labels-semantic","paper_url":"https://arxiv.org/abs/2202.02002v2","paper_title":"Scaling up Multi-domain Semantic Segmentation with Sentence Embeddings","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":24,"model":"SSDD","metrics":{"mIoU":"64.9"},"uses_additional_data":false,"paper_date":"2019-11-04","paper":"/paper/self-supervised-difference-detection-for-1","paper_url":"https://arxiv.org/abs/1911.01370v2","paper_title":"Self-Supervised Difference Detection for Weakly-Supervised Semantic Segmentation","code":"https://github.com/shimoda-uec/ssdd","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":7,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":25,"model":"PSA w/ EADER DeepLab (Xception-65)","metrics":{"mIoU":"62.8%"},"uses_additional_data":false,"paper_date":"2020-11-09","paper":"/paper/find-it-if-you-can-end-to-end-adversarial","paper_url":"https://arxiv.org/abs/2011.04626v1","paper_title":"Find it if You Can: End-to-End Adversarial Erasing for Weakly-Supervised Semantic Segmentation","code":"https://github.com/ErikStammes/EADER","n_code_links":1,"syntology":null},{"rank_in_archive_order":26,"model":"G2","metrics":{"mIoU":"55.7%"},"uses_additional_data":false,"paper_date":"2017-01-28","paper":"/paper/exploiting-saliency-for-object-segmentation","paper_url":"http://arxiv.org/abs/1701.08261v2","paper_title":"Exploiting saliency for object segmentation from image level labels","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":27,"model":"PRM","metrics":{"mIoU":"53.4%"},"uses_additional_data":false,"paper_date":"2018-04-03","paper":"/paper/weakly-supervised-instance-segmentation-using-1","paper_url":"http://arxiv.org/abs/1804.00880v1","paper_title":"Weakly Supervised Instance Segmentation using Class Peak Response","code":"https://github.com/ZhouYanzhao/PRM","n_code_links":1,"syntology":null},{"rank_in_archive_order":28,"model":"SID","metrics":{"Mean IoU":"71.6"},"uses_additional_data":false,"paper_date":"2016-03-24","paper":"/paper/simple-does-it-weakly-supervised-instance-and","paper_url":"http://arxiv.org/abs/1603.07485v2","paper_title":"Simple Does It: Weakly Supervised Instance and Semantic Segmentation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":29,"model":"DeepLabV3+ (ResNet-101)","metrics":{"mIoU (Syn)":"75.39"},"uses_additional_data":false,"paper_date":"2018-02-07","paper":"/paper/encoder-decoder-with-atrous-separable","paper_url":"http://arxiv.org/abs/1802.02611v3","paper_title":"Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation","code":"https://github.com/tensorflow/models/tree/master/research/deeplab","n_code_links":78,"syntology":{"n_ran":43,"n_unverified":29,"n_samples":72,"n_pointer_only_licence":40}}],"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":12,"rows_with_any_sample_ran":7,"distinct_papers_with_graph_line":10,"distinct_papers_with_any_sample_ran":7,"samples_over_distinct_papers":{"n_ran":81,"n_unverified":107,"n_samples":188,"n_pointer_only_licence":69,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":81,"n_unverified":135,"n_samples":216,"n_pointer_only_licence":69,"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"}}}