{"url":"/sota/panoptic-segmentation-on-mapillary-val","task":{"name":"Panoptic Segmentation","url":"/task/panoptic-segmentation","note":null},"dataset":{"name":"Mapillary val","url":"/dataset/mapillary-vistas-dataset"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"**Panoptic Segmentation** is a computer vision task that combines semantic segmentation and instance segmentation to provide a comprehensive understanding of the scene. The goal of panoptic segmentation is to segment the image into semantically meaningful parts or regions, while also detecting and distinguishing individual instances of objects within those regions. In a given image, every pixel is assigned a semantic label, and pixels belonging to \"things\" classes (countable objects with instances, like cars and people) are assigned unique instance IDs.\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [Detectron2](https://github.com/facebookresearch/detectron2) )</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":["PQ","mIoU","PQst","PQth"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"PQ":null,"mIoU":null,"PQst":null,"PQth":null}},"counts":{"rows":13,"rows_with_code":10,"rows_with_paper_page":13,"rows_dated":13,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"OneFormer (DiNAT-L, single-scale)","metrics":{"PQ":"46.7","PQst":"54.9","PQth":"40.5","mIoU":"61.7"},"uses_additional_data":false,"paper_date":"2022-11-10","paper":"/paper/oneformer-one-transformer-to-rule-universal","paper_url":"https://arxiv.org/abs/2211.06220v2","paper_title":"OneFormer: One Transformer to Rule Universal Image Segmentation","code":"https://github.com/huggingface/transformers","n_code_links":4,"syntology":{"n_ran":0,"n_unverified":5,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"OneFormer (ConvNeXt-L, single-scale)","metrics":{"PQ":"46.4","PQst":"54.0","PQth":"40.6","mIoU":"61.6"},"uses_additional_data":false,"paper_date":"2022-11-10","paper":"/paper/oneformer-one-transformer-to-rule-universal","paper_url":"https://arxiv.org/abs/2211.06220v2","paper_title":"OneFormer: One Transformer to Rule Universal Image Segmentation","code":"https://github.com/huggingface/transformers","n_code_links":4,"syntology":{"n_ran":0,"n_unverified":5,"n_samples":5,"n_pointer_only_licence":0}},{"rank_in_archive_order":3,"model":"Panoptic FCN* (Swin-L, single-scale)","metrics":{"PQ":"45.7","PQst":"52.1","PQth":"40.8"},"uses_additional_data":false,"paper_date":"2020-12-01","paper":"/paper/fully-convolutional-networks-for-panoptic","paper_url":"https://arxiv.org/abs/2012.00720v2","paper_title":"Fully Convolutional Networks for Panoptic Segmentation","code":"https://github.com/dvlab-research/panopticfcn","n_code_links":6,"syntology":null},{"rank_in_archive_order":4,"model":"Panoptic-DeepLab (SWideRNet-(1, 1, 4.5), multi-scale)","metrics":{"PQ":"44.8","PQst":"51.9","PQth":"39.3","mIoU":"60.0"},"uses_additional_data":false,"paper_date":"2020-11-23","paper":"/paper/scaling-wide-residual-networks-for-panoptic","paper_url":"https://arxiv.org/abs/2011.11675v2","paper_title":"Scaling Wide Residual Networks for Panoptic Segmentation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"Mask2Former + Intra-Batch Supervision (ResNet-50)","metrics":{"PQ":"42.2","PQst":"52.0","PQth":"34.9"},"uses_additional_data":false,"paper_date":"2023-04-17","paper":"/paper/intra-batch-supervision-for-panoptic-1","paper_url":"https://arxiv.org/abs/2304.08222v1","paper_title":"Intra-Batch Supervision for Panoptic Segmentation on High-Resolution Images","code":"https://github.com/DdeGeus/Mask2Former-IBS","n_code_links":2,"syntology":null},{"rank_in_archive_order":6,"model":"Axial-DeepLab-L (multi-scale)","metrics":{"PQ":"41.1","PQst":"51.3","PQth":"33.4","mIoU":"58.4"},"uses_additional_data":false,"paper_date":"2020-03-17","paper":"/paper/axial-deeplab-stand-alone-axial-attention-for","paper_url":"https://arxiv.org/abs/2003.07853v2","paper_title":"Axial-DeepLab: Stand-Alone Axial-Attention for Panoptic Segmentation","code":"https://github.com/The-AI-Summer/self_attention","n_code_links":5,"syntology":{"n_ran":3,"n_unverified":8,"n_samples":11,"n_pointer_only_licence":1}},{"rank_in_archive_order":7,"model":"EfficientPS","metrics":{"PQ":"40.6"},"uses_additional_data":false,"paper_date":"2020-04-05","paper":"/paper/efficientps-efficient-panoptic-segmentation","paper_url":"https://arxiv.org/abs/2004.02307v3","paper_title":"EfficientPS: Efficient Panoptic Segmentation","code":"https://github.com/DeepSceneSeg/EfficientPS","n_code_links":2,"syntology":null},{"rank_in_archive_order":8,"model":"Panoptic-DeepLab (X71)","metrics":{"PQ":"40.5"},"uses_additional_data":false,"paper_date":"2019-11-22","paper":"/paper/panoptic-deeplab-a-simple-strong-and-fast","paper_url":"https://arxiv.org/abs/1911.10194v3","paper_title":"Panoptic-DeepLab: A Simple, Strong, and Fast Baseline for Bottom-Up Panoptic Segmentation","code":"https://github.com/tensorflow/models/tree/master/official/projects/panoptic","n_code_links":9,"syntology":{"n_ran":3,"n_unverified":6,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"AdaptIS (ResNeXt-101)","metrics":{"PQ":"40.3","mIoU":"56.8"},"uses_additional_data":false,"paper_date":"2019-09-17","paper":"/paper/adaptis-adaptive-instance-selection-network","paper_url":"https://arxiv.org/abs/1909.07829v1","paper_title":"AdaptIS: Adaptive Instance Selection Network","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":10,"model":"Panoptic FCN* (ResNet-FPN)","metrics":{"PQ":"36.9","PQth":"32.9"},"uses_additional_data":false,"paper_date":"2020-12-01","paper":"/paper/fully-convolutional-networks-for-panoptic","paper_url":"https://arxiv.org/abs/2012.00720v2","paper_title":"Fully Convolutional Networks for Panoptic Segmentation","code":"https://github.com/dvlab-research/panopticfcn","n_code_links":6,"syntology":null},{"rank_in_archive_order":11,"model":"JSIS-Net (ResNet-50)","metrics":{"PQ":"17.6"},"uses_additional_data":false,"paper_date":"2018-09-06","paper":"/paper/panoptic-segmentation-with-a-joint-semantic","paper_url":"http://arxiv.org/abs/1809.02110v2","paper_title":"Panoptic Segmentation with a Joint Semantic and Instance Segmentation Network","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":12,"model":"HRNet-OCR (Hierarchical Multi-Scale Attention)","metrics":{"PQ":"17.6"},"uses_additional_data":false,"paper_date":"2020-05-21","paper":"/paper/hierarchical-multi-scale-attention-for","paper_url":"https://arxiv.org/abs/2005.10821v1","paper_title":"Hierarchical Multi-Scale Attention for Semantic Segmentation","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":8,"syntology":{"n_ran":1,"n_unverified":8,"n_samples":9,"n_pointer_only_licence":1}},{"rank_in_archive_order":13,"model":"Panoptic FCN* (ResNet-50-FPN)","metrics":{"PQst":"42.3"},"uses_additional_data":false,"paper_date":"2020-12-01","paper":"/paper/fully-convolutional-networks-for-panoptic","paper_url":"https://arxiv.org/abs/2012.00720v2","paper_title":"Fully Convolutional Networks for Panoptic Segmentation","code":"https://github.com/dvlab-research/panopticfcn","n_code_links":6,"syntology":null}],"since_archive":{"present":false,"note":"No Syntology-extracted rows are published in this build."},"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":3,"distinct_papers_with_graph_line":4,"distinct_papers_with_any_sample_ran":3,"samples_over_distinct_papers":{"n_ran":7,"n_unverified":27,"n_samples":34,"n_pointer_only_licence":2,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":7,"n_unverified":32,"n_samples":39,"n_pointer_only_licence":2,"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"}}}