{"url":"/sota/real-time-semantic-segmentation-on-cityscapes-1","task":{"name":"Real-Time Semantic Segmentation","url":"/task/real-time-semantic-segmentation","note":null},"dataset":{"name":"Cityscapes val","url":"/dataset/cityscapes"},"category":"Computer Vision","categories":["Computer Code","Computer Vision","Medical","Robots"],"category_note":null,"description":"Semantic Segmentation is a computer vision task that involves assigning a semantic label to each pixel in an image. In **Real-Time Semantic Segmentation**, the goal is to perform this labeling quickly and accurately in real-time, allowing for the segmentation results to be used for tasks such as object recognition, scene understanding, and autonomous navigation.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [TorchSeg](https://github.com/ycszen/TorchSeg) )</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","Frame (fps)","Time (ms)"],"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,"Frame (fps)":null,"Time (ms)":"lower"}},"counts":{"rows":24,"rows_with_code":24,"rows_with_paper_page":24,"rows_dated":24,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"PIDNet-L","metrics":{"Frame (fps)":"31.1(3090)","Time (ms)":"32.2","mIoU":"80.9%"},"uses_additional_data":false,"paper_date":"2022-06-04","paper":"/paper/pidnet-a-real-time-semantic-segmentation","paper_url":"https://arxiv.org/abs/2206.02066v3","paper_title":"PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers","code":"https://github.com/XuJiacong/PIDNet","n_code_links":6,"syntology":{"n_ran":9,"n_unverified":10,"n_samples":19,"n_pointer_only_licence":1}},{"rank_in_archive_order":2,"model":"DSNet","metrics":{"Frame (fps)":"81.9","mIoU":"80.4%"},"uses_additional_data":false,"paper_date":"2024-06-06","paper":"/paper/dsnet-a-novel-way-to-use-atrous-convolutions","paper_url":"https://arxiv.org/abs/2406.03702v1","paper_title":"DSNet: A Novel Way to Use Atrous Convolutions in Semantic Segmentation","code":"https://github.com/takaniwa/dsnet","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"PIDNet-M","metrics":{"Frame (fps)":"42.2(3090)","Time (ms)":"23.7","mIoU":"79.9%"},"uses_additional_data":false,"paper_date":"2022-06-04","paper":"/paper/pidnet-a-real-time-semantic-segmentation","paper_url":"https://arxiv.org/abs/2206.02066v3","paper_title":"PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers","code":"https://github.com/XuJiacong/PIDNet","n_code_links":6,"syntology":{"n_ran":9,"n_unverified":10,"n_samples":19,"n_pointer_only_licence":1}},{"rank_in_archive_order":4,"model":"SegNext-T-Seg100","metrics":{"Frame (fps)":"28.1","mIoU":"79.8%"},"uses_additional_data":false,"paper_date":"2022-09-18","paper":"/paper/segnext-rethinking-convolutional-attention","paper_url":"https://arxiv.org/abs/2209.08575v1","paper_title":"SegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation","code":"https://github.com/open-mmlab/mmsegmentation","n_code_links":5,"syntology":null},{"rank_in_archive_order":5,"model":"DDRNet23","metrics":{"Frame (fps)":"37.1","mIoU":"79.4"},"uses_additional_data":false,"paper_date":"2021-01-15","paper":"/paper/deep-dual-resolution-networks-for-real-time","paper_url":"https://arxiv.org/abs/2101.06085v2","paper_title":"Deep Dual-resolution Networks for Real-time and Accurate Semantic Segmentation of Road Scenes","code":"https://github.com/Deci-AI/super-gradients","n_code_links":8,"syntology":null},{"rank_in_archive_order":6,"model":"RTFormer-B","metrics":{"Frame (fps)":"50.2","mIoU":"79.3%"},"uses_additional_data":false,"paper_date":"2022-10-13","paper":"/paper/rtformer-efficient-design-for-real-time","paper_url":"https://arxiv.org/abs/2210.07124v1","paper_title":"RTFormer: Efficient Design for Real-Time Semantic Segmentation with Transformer","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":1,"syntology":null},{"rank_in_archive_order":7,"model":"PIDNet-S","metrics":{"Frame (fps)":"93.2(3090)","Time (ms)":"10.7","mIoU":"78.8%"},"uses_additional_data":false,"paper_date":"2022-06-04","paper":"/paper/pidnet-a-real-time-semantic-segmentation","paper_url":"https://arxiv.org/abs/2206.02066v3","paper_title":"PIDNet: A Real-time Semantic Segmentation Network Inspired by PID Controllers","code":"https://github.com/XuJiacong/PIDNet","n_code_links":6,"syntology":{"n_ran":9,"n_unverified":10,"n_samples":19,"n_pointer_only_licence":1}},{"rank_in_archive_order":8,"model":"Mobile-Seed","metrics":{"Frame (fps)":"23.9(2080Ti)","mIoU":"78.4%"},"uses_additional_data":false,"paper_date":"2023-11-21","paper":"/paper/mobile-seed-joint-semantic-segmentation-and","paper_url":"https://arxiv.org/abs/2311.12651v3","paper_title":"Mobile-Seed: Joint Semantic Segmentation and Boundary Detection for Mobile Robots","code":"https://github.com/whu-usi3dv/mobile-seed","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"PP-LiteSeg-B2","metrics":{"mIoU":"78.2"},"uses_additional_data":false,"paper_date":"2022-04-06","paper":"/paper/pp-liteseg-a-superior-real-time-semantic","paper_url":"https://arxiv.org/abs/2204.02681v1","paper_title":"PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":3,"syntology":null},{"rank_in_archive_order":10,"model":"DDRNet23-slim","metrics":{"Frame (fps)":"101.6","mIoU":"77.4"},"uses_additional_data":false,"paper_date":"2021-01-15","paper":"/paper/deep-dual-resolution-networks-for-real-time","paper_url":"https://arxiv.org/abs/2101.06085v2","paper_title":"Deep Dual-resolution Networks for Real-time and Accurate Semantic Segmentation of Road Scenes","code":"https://github.com/Deci-AI/super-gradients","n_code_links":8,"syntology":null},{"rank_in_archive_order":11,"model":"STDC2-Seg75","metrics":{"Frame (fps)":"97","mIoU":"77%"},"uses_additional_data":false,"paper_date":"2021-04-27","paper":"/paper/rethinking-bisenet-for-real-time-semantic","paper_url":"https://arxiv.org/abs/2104.13188v1","paper_title":"Rethinking BiSeNet For Real-time Semantic Segmentation","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":6,"syntology":{"n_ran":9,"n_unverified":7,"n_samples":16,"n_pointer_only_licence":0}},{"rank_in_archive_order":12,"model":"CSFNet-2","metrics":{"Frame (fps)":"72.3 (3090)","mIoU":"76.36"},"uses_additional_data":false,"paper_date":"2024-07-01","paper":"/paper/csfnet-a-cosine-similarity-fusion-network-for","paper_url":"https://arxiv.org/abs/2407.01328v1","paper_title":"CSFNet: A Cosine Similarity Fusion Network for Real-Time RGB-X Semantic Segmentation of Driving Scenes","code":"https://github.com/Danial-Qashqai/CSFNet","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"RTFormer-S","metrics":{"Frame (fps)":"89.6","mIoU":"76.3%"},"uses_additional_data":false,"paper_date":"2022-10-13","paper":"/paper/rtformer-efficient-design-for-real-time","paper_url":"https://arxiv.org/abs/2210.07124v1","paper_title":"RTFormer: Efficient Design for Real-Time Semantic Segmentation with Transformer","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"PP-LiteSeg-T2","metrics":{"mIoU":"76"},"uses_additional_data":false,"paper_date":"2022-04-06","paper":"/paper/pp-liteseg-a-superior-real-time-semantic","paper_url":"https://arxiv.org/abs/2204.02681v1","paper_title":"PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":3,"syntology":null},{"rank_in_archive_order":15,"model":"BiseNetV2-L","metrics":{"Frame (fps)":"47.3","mIoU":"75.8%"},"uses_additional_data":false,"paper_date":"2020-04-05","paper":"/paper/bisenet-v2-bilateral-network-with-guided","paper_url":"https://arxiv.org/abs/2004.02147v1","paper_title":"BiSeNet V2: Bilateral Network with Guided Aggregation for Real-time Semantic Segmentation","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":7,"syntology":{"n_ran":7,"n_unverified":3,"n_samples":10,"n_pointer_only_licence":5}},{"rank_in_archive_order":16,"model":"SwiftNetRN-18","metrics":{"Frame (fps)":"39.9","mIoU":"75.5%"},"uses_additional_data":false,"paper_date":"2020-04-24","paper":"/paper/in-defence-of-metric-learning-for-speaker","paper_url":"http://arxiv.org/abs/2003.11982v2","paper_title":"In defence of metric learning for speaker recognition","code":"https://github.com/coqui-ai/TTS","n_code_links":4,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":1}},{"rank_in_archive_order":17,"model":"PP-LiteSeg-B1","metrics":{"mIoU":"75.3"},"uses_additional_data":false,"paper_date":"2022-04-06","paper":"/paper/pp-liteseg-a-superior-real-time-semantic","paper_url":"https://arxiv.org/abs/2204.02681v1","paper_title":"PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":3,"syntology":null},{"rank_in_archive_order":18,"model":"CSFNet-1","metrics":{"Frame (fps)":"106.1","mIoU":"74.73"},"uses_additional_data":false,"paper_date":"2024-07-01","paper":"/paper/csfnet-a-cosine-similarity-fusion-network-for","paper_url":"https://arxiv.org/abs/2407.01328v1","paper_title":"CSFNet: A Cosine Similarity Fusion Network for Real-Time RGB-X Semantic Segmentation of Driving Scenes","code":"https://github.com/Danial-Qashqai/CSFNet","n_code_links":1,"syntology":null},{"rank_in_archive_order":19,"model":"STDC1-Seg75","metrics":{"Frame (fps)":"126.7","mIoU":"74.5%"},"uses_additional_data":false,"paper_date":"2021-04-27","paper":"/paper/rethinking-bisenet-for-real-time-semantic","paper_url":"https://arxiv.org/abs/2104.13188v1","paper_title":"Rethinking BiSeNet For Real-time Semantic Segmentation","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":6,"syntology":{"n_ran":9,"n_unverified":7,"n_samples":16,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"BiseNetV2","metrics":{"Frame (fps)":"156","mIoU":"73.5%"},"uses_additional_data":false,"paper_date":"2020-04-05","paper":"/paper/bisenet-v2-bilateral-network-with-guided","paper_url":"https://arxiv.org/abs/2004.02147v1","paper_title":"BiSeNet V2: Bilateral Network with Guided Aggregation for Real-time Semantic Segmentation","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":7,"syntology":{"n_ran":7,"n_unverified":3,"n_samples":10,"n_pointer_only_licence":5}},{"rank_in_archive_order":21,"model":"FasterSeg","metrics":{"Frame (fps)":"163.9","mIoU":"73.1"},"uses_additional_data":false,"paper_date":"2019-12-23","paper":"/paper/fasterseg-searching-for-faster-real-time-1","paper_url":"https://arxiv.org/abs/1912.10917v2","paper_title":"FasterSeg: Searching for Faster Real-time Semantic Segmentation","code":"https://github.com/TAMU-VITA/FasterSeg","n_code_links":2,"syntology":{"n_ran":7,"n_unverified":6,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"PP-LiteSeg-T1","metrics":{"mIoU":"73.1"},"uses_additional_data":false,"paper_date":"2022-04-06","paper":"/paper/pp-liteseg-a-superior-real-time-semantic","paper_url":"https://arxiv.org/abs/2204.02681v1","paper_title":"PP-LiteSeg: A Superior Real-Time Semantic Segmentation Model","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":3,"syntology":null},{"rank_in_archive_order":23,"model":"LDFNet","metrics":{"mIoU":"68.48%"},"uses_additional_data":false,"paper_date":"2018-09-24","paper":"/paper/incorporating-luminance-depth-and-color","paper_url":"https://arxiv.org/abs/1809.09077v3","paper_title":"Incorporating Luminance, Depth and Color Information by a Fusion-based Network for Semantic Segmentation","code":"https://github.com/shangweihung/LDFNet","n_code_links":1,"syntology":null},{"rank_in_archive_order":24,"model":"LiteSeg-MobileNet","metrics":{"mIoU":"67.8%"},"uses_additional_data":false,"paper_date":"2019-12-13","paper":"/paper/liteseg-a-novel-lightweight-convnet-for","paper_url":"https://arxiv.org/abs/1912.06683v1","paper_title":"LiteSeg: A Novel Lightweight ConvNet for Semantic Segmentation","code":"https://github.com/zh320/realtime-semantic-segmentation-pytorch","n_code_links":2,"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 6,885 of the 9,623 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":9623,"papers_checked":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+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":9,"rows_with_any_sample_ran":9,"distinct_papers_with_graph_line":5,"distinct_papers_with_any_sample_ran":5,"samples_over_distinct_papers":{"n_ran":35,"n_unverified":26,"n_samples":61,"n_pointer_only_licence":7,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":69,"n_unverified":56,"n_samples":125,"n_pointer_only_licence":14,"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"}}}