{"url":"/sota/real-time-instance-segmentation-on-mscoco","task":{"name":"Real-time Instance Segmentation","url":"/task/real-time-instance-segmentation","note":null},"dataset":{"name":"MSCOCO","url":"/dataset/mscoco"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"Similar to its parent task, instance segmentation, but with the goal of achieving real-time capabilities under a defined setting.\r\n\r\nImage Credit: [SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation](https://arxiv.org/pdf/2007.14772v1.pdf)","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":["mask AP","AP50","AP75","APS","APM","APL","Frame (fps)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"mask AP":"higher","AP50":null,"AP75":null,"APS":null,"APM":null,"APL":null,"Frame (fps)":null}},"counts":{"rows":22,"rows_with_code":17,"rows_with_paper_page":17,"rows_dated":17,"rows_using_additional_data":2},"rows":[{"rank_in_archive_order":1,"model":"RTMDet-Ins-x","metrics":{"AP50":"67.4","AP75":"47.8","APL":"65.5","APS":"22.2","Frame (fps)":"188\n(RTX3090)","mask AP":"44.6"},"uses_additional_data":false,"paper_date":"2022-12-14","paper":"/paper/rtmdet-an-empirical-study-of-designing-real","paper_url":"https://arxiv.org/abs/2212.07784v2","paper_title":"RTMDet: An Empirical Study of Designing Real-Time Object Detectors","code":"https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet","n_code_links":14,"syntology":{"n_ran":3,"n_unverified":17,"n_samples":20,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"RTMDet-Ins-l","metrics":{"AP50":"66","AP75":"47.0","APL":"64.8","APM":"48.0","APS":"20.8","Frame (fps)":"271\n(RTX3090)","mask AP":"43.7"},"uses_additional_data":false,"paper_date":"2022-12-14","paper":"/paper/rtmdet-an-empirical-study-of-designing-real","paper_url":"https://arxiv.org/abs/2212.07784v2","paper_title":"RTMDet: An Empirical Study of Designing Real-Time Object Detectors","code":"https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet","n_code_links":14,"syntology":{"n_ran":3,"n_unverified":17,"n_samples":20,"n_pointer_only_licence":0}},{"rank_in_archive_order":3,"model":"RTMDet-Ins-m","metrics":{"AP50":"63.9","AP75":"45.1","APL":"63.1","APM":"46.4","APS":"19.3","Frame (fps)":"371\n(RTX3090)","mask AP":"42.1"},"uses_additional_data":false,"paper_date":"2022-12-14","paper":"/paper/rtmdet-an-empirical-study-of-designing-real","paper_url":"https://arxiv.org/abs/2212.07784v2","paper_title":"RTMDet: An Empirical Study of Designing Real-Time Object Detectors","code":"https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet","n_code_links":14,"syntology":{"n_ran":3,"n_unverified":17,"n_samples":20,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"RetinaMask","metrics":{"Frame (fps)":"10","mask AP":"41.4"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"Predicting Future Instance Segmentation","metrics":{"Frame (fps)":"8","mask AP":"40.2"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":6,"model":"UOAIS","metrics":{"Frame (fps)":"12","mask AP":"39.6"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":7,"model":"SparseInst-608 (ResNet-50-vd)","metrics":{"AP50":"59.2","AP75":"40.2","APL":"56.9","APM":"39.4","APS":"15.7","Frame (fps)":"40 (2080 Ti)","mask AP":"37.9"},"uses_additional_data":false,"paper_date":"2022-03-24","paper":"/paper/sparse-instance-activation-for-real-time","paper_url":"https://arxiv.org/abs/2203.12827v1","paper_title":"Sparse Instance Activation for Real-Time Instance Segmentation","code":"https://github.com/hustvl/sparseinst","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":8,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":8,"model":"maYOLACT-700 (ResNet-50)","metrics":{"AP50":"59.4","AP75":"39.9","APL":"52.5","APM":"40.8","APS":"18.1","Frame (fps)":"25 (Tesla V100)","mask AP":"37.7"},"uses_additional_data":false,"paper_date":"2021-10-19","paper":"/paper/mask-aware-iou-for-anchor-assignment-in-real","paper_url":"https://arxiv.org/abs/2110.09734v1","paper_title":"Mask-aware IoU for Anchor Assignment in Real-time Instance Segmentation","code":"https://github.com/kemaloksuz/mask-aware-iou","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"SOLO-512","metrics":{"AP50":"57.7","AP75":"39.7","Frame (fps)":"31.3","mask AP":"37.1"},"uses_additional_data":false,"paper_date":"2020-03-23","paper":"/paper/solov2-dynamic-faster-and-stronger","paper_url":"https://arxiv.org/abs/2003.10152v3","paper_title":"SOLOv2: Dynamic and Fast Instance Segmentation","code":"https://github.com/open-mmlab/mmdetection","n_code_links":18,"syntology":{"n_ran":15,"n_unverified":23,"n_samples":38,"n_pointer_only_licence":24}},{"rank_in_archive_order":10,"model":"SparseInst-448 (ResNet-50-vd)","metrics":{"AP50":"56.5","AP75":"37.7","APL":"57.0","APM":"37.1","APS":"12.3","Frame (fps)":"58.5 (2080 Ti)","mask AP":"35.9"},"uses_additional_data":false,"paper_date":"2022-03-24","paper":"/paper/sparse-instance-activation-for-real-time","paper_url":"https://arxiv.org/abs/2203.12827v1","paper_title":"Sparse Instance Activation for Real-Time Instance Segmentation","code":"https://github.com/hustvl/sparseinst","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":8,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"SipMask++ (ResNet-101, single-scale test)","metrics":{"AP50":"55.6","AP75":"37.6","APL":"56.8","APM":"38.3","APS":"11.2","Frame (fps)":"27.0 (Titan Xp)","mask AP":"35.4"},"uses_additional_data":false,"paper_date":"2020-07-29","paper":"/paper/sipmask-spatial-information-preservation-for","paper_url":"https://arxiv.org/abs/2007.14772v1","paper_title":"SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation","code":"https://github.com/JialeCao001/SipMask","n_code_links":1,"syntology":null},{"rank_in_archive_order":12,"model":"maYOLACT-550 (ResNet-50)","metrics":{"AP50":"56.2","AP75":"37.1","APL":"51.4","APM":"38.0","APS":"14.7","Frame (fps)":"30 (Tesla V100)","mask AP":"35.2"},"uses_additional_data":false,"paper_date":"2021-10-19","paper":"/paper/mask-aware-iou-for-anchor-assignment-in-real","paper_url":"https://arxiv.org/abs/2110.09734v1","paper_title":"Mask-aware IoU for Anchor Assignment in Real-time Instance Segmentation","code":"https://github.com/kemaloksuz/mask-aware-iou","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"BlendMask-512 (DLA_34)","metrics":{"Frame (fps)":"33.3","mask AP":"35.2"},"uses_additional_data":false,"paper_date":"2020-01-02","paper":"/paper/blendmask-top-down-meets-bottom-up-for","paper_url":"https://arxiv.org/abs/2001.00309v3","paper_title":"BlendMask: Top-Down Meets Bottom-Up for Instance Segmentation","code":"https://github.com/aim-uofa/AdelaiDet","n_code_links":9,"syntology":null},{"rank_in_archive_order":14,"model":"YOLACT-550++ (ResNet-101-FPN)","metrics":{"AP50":"53.8","AP75":"36.9","APL":"55.1","APM":"36.8","APS":"11.9","Frame (fps)":"27.3 (Titan Xp)","mask AP":"34.6"},"uses_additional_data":false,"paper_date":"2019-12-03","paper":"/paper/yolact-better-real-time-instance-segmentation","paper_url":"https://arxiv.org/abs/1912.06218v2","paper_title":"YOLACT++: Better Real-time Instance Segmentation","code":"https://github.com/dbolya/yolact","n_code_links":36,"syntology":{"n_ran":11,"n_unverified":32,"n_samples":43,"n_pointer_only_licence":0}},{"rank_in_archive_order":15,"model":"Mask R-CNN","metrics":{"Frame (fps)":"5","mask AP":"34.6"},"uses_additional_data":true,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"CenterMask-Lite (ResNet-50-FPN)","metrics":{"APL":"48.7","APM":"34.7","APS":"12.9","mask AP":"32.9"},"uses_additional_data":false,"paper_date":"2019-11-15","paper":"/paper/centermask-real-time-anchor-free-instance-1","paper_url":"https://arxiv.org/abs/1911.06667v6","paper_title":"CenterMask : Real-Time Anchor-Free Instance Segmentation","code":"https://github.com/youngwanLEE/centermask2","n_code_links":8,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":17,"model":"SipMask (ResNet-101, single-scale test)","metrics":{"AP50":"53.4","AP75":"34.3","APL":"54.0","APM":"35.6","APS":"9.3","Frame (fps)":"31.3 (Titan Xp)","mask AP":"32.8"},"uses_additional_data":false,"paper_date":"2020-07-29","paper":"/paper/sipmask-spatial-information-preservation-for","paper_url":"https://arxiv.org/abs/2007.14772v1","paper_title":"SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation","code":"https://github.com/JialeCao001/SipMask","n_code_links":1,"syntology":null},{"rank_in_archive_order":18,"model":"SipMask (ResNet-50, single-scale test)","metrics":{"AP50":"51.9","AP75":"32.3","APL":"49.8","APM":"33.6","APS":"9.2","Frame (fps)":"41.7 (Titan Xp)","mask AP":"31.2"},"uses_additional_data":false,"paper_date":"2020-07-29","paper":"/paper/sipmask-spatial-information-preservation-for","paper_url":"https://arxiv.org/abs/2007.14772v1","paper_title":"SipMask: Spatial Information Preservation for Fast Image and Video Instance Segmentation","code":"https://github.com/JialeCao001/SipMask","n_code_links":1,"syntology":null},{"rank_in_archive_order":19,"model":"YolactEdge","metrics":{"Frame (fps)":"172.7","mask AP":"29.5"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":20,"model":"YOLACT-550 (ResNet-101-FPN)","metrics":{"AP50":"46.6","AP75":"29.2","APL":"44.8","APM":"29.3","APS":"9.2","Frame (fps)":"33.3 (Titan Xp)","mask AP":"28.2"},"uses_additional_data":false,"paper_date":"2019-04-04","paper":"/paper/yolact-real-time-instance-segmentation","paper_url":"https://arxiv.org/abs/1904.02689v2","paper_title":"YOLACT: Real-time Instance Segmentation","code":"https://github.com/open-mmlab/mmdetection","n_code_links":48,"syntology":{"n_ran":10,"n_unverified":11,"n_samples":21,"n_pointer_only_licence":6}},{"rank_in_archive_order":21,"model":"YOLACT","metrics":{"AP50":"42.0","AP75":"25.4","APL":"45.0","APM":"25.3","APS":"5.0","Frame (fps)":"45.3 (Titan Xp)","mask AP":"24.9"},"uses_additional_data":true,"paper_date":"2019-04-04","paper":"/paper/yolact-real-time-instance-segmentation","paper_url":"https://arxiv.org/abs/1904.02689v2","paper_title":"YOLACT: Real-time Instance Segmentation","code":"https://github.com/open-mmlab/mmdetection","n_code_links":48,"syntology":{"n_ran":10,"n_unverified":11,"n_samples":21,"n_pointer_only_licence":6}},{"rank_in_archive_order":22,"model":"RTMDet-Ins-s","metrics":{"AP50":"59.3","AP75":"41.3","APL":"60.3","APM":"42.3","APS":"15.1","Frame (fps)":"518\n(RTX3090)"},"uses_additional_data":false,"paper_date":"2022-12-14","paper":"/paper/rtmdet-an-empirical-study-of-designing-real","paper_url":"https://arxiv.org/abs/2212.07784v2","paper_title":"RTMDet: An Empirical Study of Designing Real-Time Object Detectors","code":"https://github.com/open-mmlab/mmdetection/tree/3.x/configs/rtmdet","n_code_links":14,"syntology":{"n_ran":3,"n_unverified":17,"n_samples":20,"n_pointer_only_licence":0}}],"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":11,"rows_with_any_sample_ran":11,"distinct_papers_with_graph_line":6,"distinct_papers_with_any_sample_ran":6,"samples_over_distinct_papers":{"n_ran":43,"n_unverified":92,"n_samples":135,"n_pointer_only_licence":32,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":65,"n_unverified":162,"n_samples":227,"n_pointer_only_licence":38,"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"}}}