{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/sparse-r-cnn-end-to-end-object-detection-with","title":"Sparse R-CNN: End-to-End Object Detection with Learnable Proposals","arxiv_id":"2011.12450","date":"2020-11-25","proceeding":"CVPR 2021 1","authors":["Peize Sun","Rufeng Zhang","Yi Jiang","Tao Kong","Chenfeng Xu","Wei Zhan","Masayoshi Tomizuka","Lei LI","Zehuan Yuan","Changhu Wang","Ping Luo"],"abstract":"We present Sparse R-CNN, a purely sparse method for object detection in images. Existing works on object detection heavily rely on dense object candidates, such as $k$ anchor boxes pre-defined on all grids of image feature map of size $H\\times W$. In our method, however, a fixed sparse set of learned object proposals, total length of $N$, are provided to object recognition head to perform classification and location. By eliminating $HWk$ (up to hundreds of thousands) hand-designed object candidates to $N$ (e.g. 100) learnable proposals, Sparse R-CNN completely avoids all efforts related to object candidates design and many-to-one label assignment. More importantly, final predictions are directly output without non-maximum suppression post-procedure. Sparse R-CNN demonstrates accuracy, run-time and training convergence performance on par with the well-established detector baselines on the challenging COCO dataset, e.g., achieving 45.0 AP in standard $3\\times$ training schedule and running at 22 fps using ResNet-50 FPN model. We hope our work could inspire re-thinking the convention of dense prior in object detectors. The code is available at: https://github.com/PeizeSun/SparseR-CNN.","url_abs":"https://arxiv.org/abs/2011.12450v2","url_pdf":"https://arxiv.org/pdf/2011.12450v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"sparse-r-cnn-end-to-end-object-detection-with","repo_url":"https://github.com/PeizeSun/SparseR-CNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"sparse-r-cnn-end-to-end-object-detection-with","repo_url":"https://github.com/Booomshaker/SparseRCNNWSL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"sparse-r-cnn-end-to-end-object-detection-with","repo_url":"https://github.com/henbucuoshanghai/sparsercnn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"sparse-r-cnn-end-to-end-object-detection-with","repo_url":"https://github.com/liangheming/sparse_rcnnv1","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"sparse-r-cnn-end-to-end-object-detection-with","repo_url":"https://github.com/PaddlePaddle/PaddleDetection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"paddle","reach":null},{"paper_slug":"sparse-r-cnn-end-to-end-object-detection-with","repo_url":"https://github.com/open-mmlab/mmdetection","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"2d-object-detection","task_name":"2D Object Detection"},{"task_slug":"object","task_name":"Object"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"object-recognition","task_name":"Object Recognition"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"fpn","method_name":"FPN"},{"method_slug":"sparse-r-cnn","method_name":"Sparse R-CNN"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/2d-object-detection-on-ceymo","task":"2D Object Detection","dataset":"CeyMo","model":"Sparse R-CNN","rank_in_archive_order":5,"of":5,"metrics":{"mAP":"47.3"},"uses_additional_data":false},{"leaderboard":"/sota/2d-object-detection-on-sardet-100k","task":"2D Object Detection","dataset":"SARDet-100K","model":"Sparse R-CNN","rank_in_archive_order":12,"of":13,"metrics":{"box mAP":"38.1"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-minival","task":"Object Detection","dataset":"COCO minival","model":"Sparse R-CNN (ResNet-101, learnable proposals, random crop aug, FPN)","rank_in_archive_order":112,"of":220,"metrics":{"AP50":"64.6","AP75":"49.5","APL":"61.6","APM":"48.3","APS":"28.3","box AP":"45.6"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-minival","task":"Object Detection","dataset":"COCO minival","model":"Sparse R-CNN (ResNet-50, learnable proposals, random crop aug, FPN)","rank_in_archive_order":128,"of":220,"metrics":{"AP50":"63.4","AP75":"48.2","APL":"59.5","APM":"47.2","APS":"26.9","box AP":"44.5"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-minival","task":"Object Detection","dataset":"COCO minival","model":"Sparse R-CNN (ResNet-101, FPN)","rank_in_archive_order":138,"of":220,"metrics":{"AP50":"62.1","AP75":"47.2","APL":"59.7","APM":"46.3","APS":"26.1","box AP":"43.5"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-minival","task":"Object Detection","dataset":"COCO minival","model":"Sparse R-CNN (ResNet-50, FPN)","rank_in_archive_order":155,"of":220,"metrics":{"AP50":"61.2","AP75":"45.7","APL":"57.6","APM":"44.6","APS":"26.7","box AP":"42.3"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2011.12450","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}