{"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/relation-detr-exploring-explicit-position","title":"Relation DETR: Exploring Explicit Position Relation Prior for Object Detection","arxiv_id":"2407.11699","date":"2024-07-16","proceeding":null,"authors":["Xiuquan Hou","Meiqin Liu","Senlin Zhang","Ping Wei","Badong Chen","Xuguang Lan"],"abstract":"This paper presents a general scheme for enhancing the convergence and performance of DETR (DEtection TRansformer). We investigate the slow convergence problem in transformers from a new perspective, suggesting that it arises from the self-attention that introduces no structural bias over inputs. To address this issue, we explore incorporating position relation prior as attention bias to augment object detection, following the verification of its statistical significance using a proposed quantitative macroscopic correlation (MC) metric. Our approach, termed Relation-DETR, introduces an encoder to construct position relation embeddings for progressive attention refinement, which further extends the traditional streaming pipeline of DETR into a contrastive relation pipeline to address the conflicts between non-duplicate predictions and positive supervision. Extensive experiments on both generic and task-specific datasets demonstrate the effectiveness of our approach. Under the same configurations, Relation-DETR achieves a significant improvement (+2.0% AP compared to DINO), state-of-the-art performance (51.7% AP for 1x and 52.1% AP for 2x settings), and a remarkably faster convergence speed (over 40% AP with only 2 training epochs) than existing DETR detectors on COCO val2017. Moreover, the proposed relation encoder serves as a universal plug-in-and-play component, bringing clear improvements for theoretically any DETR-like methods. Furthermore, we introduce a class-agnostic detection dataset, SA-Det-100k. The experimental results on the dataset illustrate that the proposed explicit position relation achieves a clear improvement of 1.3% AP, highlighting its potential towards universal object detection. The code and dataset are available at https://github.com/xiuqhou/Relation-DETR.","url_abs":"https://arxiv.org/abs/2407.11699v1","url_pdf":"https://arxiv.org/pdf/2407.11699v1.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":"relation-detr-exploring-explicit-position","repo_url":"https://github.com/xiuqhou/relation-detr","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"relation-detr-exploring-explicit-position","repo_url":"https://github.com/xiuqhou/Salience-DETR","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"2d-object-detection","task_name":"2D Object Detection"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":null,"task_name":"Position"},{"task_slug":null,"task_name":"Relation"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"detr","method_name":"Detr"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"feedforward-network","method_name":"Feedforward Network"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"speed","method_name":"SPEED"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[{"slug":"sa-det-100k","name":"SA-Det-100k","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/object-detection-on-coco-2017-val","task":"Object Detection","dataset":"COCO 2017 val","model":"Relation-DETR (Swin-L 2x)","rank_in_archive_order":4,"of":33,"metrics":{"AP":"58.1","AP50":"76.4","AP75":"63.5","APL":"73.5","APM":"63.0","APS":"41.8"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-2017-val","task":"Object Detection","dataset":"COCO 2017 val","model":"Relation-DETR (Swin-L 1x)","rank_in_archive_order":5,"of":33,"metrics":{"AP":"57.8","AP50":"76.1","AP75":"62.9","APL":"74.4","APM":"62.1","APS":"41.2"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-2017-val","task":"Object Detection","dataset":"COCO 2017 val","model":"Relation-DETR (ResNet50 2x)","rank_in_archive_order":12,"of":33,"metrics":{"AP":"52.1","AP50":"69.7","AP75":"56.6","APL":"66.5","APM":"56.0","APS":"36.1"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco-2017-val","task":"Object Detection","dataset":"COCO 2017 val","model":"Relation-DETR (ResNet50 1x)","rank_in_archive_order":13,"of":33,"metrics":{"AP":"51.7","AP50":"69.1","AP75":"56.3","APL":"66.1","APM":"55.6","APS":"36.1"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-coco","task":"Object Detection","dataset":"COCO test-dev","model":"Relation-DETR (Focal-L)","rank_in_archive_order":15,"of":225,"metrics":{"AP50":"80.8","AP75":"69.1","APL":"77.0","APM":"66.9","APS":"47.2","Params (M)":"214","box mAP":"63.5"},"uses_additional_data":false},{"leaderboard":"/sota/object-detection-on-sa-det-100k","task":"Object Detection","dataset":"SA-Det-100k","model":"Relation-DETR (ResNet50 1x)","rank_in_archive_order":1,"of":2,"metrics":{"AP":"45.0","AP50":"53.1","AP75":"48.9","APL":"62.9","APM":"44.4","APS":"6.0"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2407.11699","mcp":{"get_harvested_code_for_paper":{"arxiv_id":"2407.11699"}},"developers":"https://syntology.ai/developers","read_at":"2026-09-24T18:15:14+00:00","read_at_is":"when the build read Syntology's graph, not when any sample ran","claim":"Per-sample execution status on synthesized fixtures; not a correctness claim about the paper. Samples come from repositories linked to the paper, official or community; repo_kind says which.","repos":[{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/xiuqhou/Salience-DETR","reach":null},{"provenance":"external:paperswithcode_snapshot_2025-07-28","url":"https://github.com/xiuqhou/relation-detr","reach":null}],"summary":{"ran_violates":1},"by_repo_kind":{"official":{"samples":1,"ran":1,"repositories":1}},"repo_kind_vocabulary":{"official":"The archive marks this repository official for the paper","named_in_paper":"The archive records that the paper mentions this repository; it is not marked official","listed":"In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper","found_in_text":"Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted","community":"Not in the archive's code links for this paper; a community repository Syntology harvested"},"n_pointer_only_for_licence":0,"samples":[{"code_sha256_prefix":"fe838c2ebeafd791","entry":"is_image","repo":"xiuqhou/relation-detr","repo_kind":"official","path":"inference.py","file_url":"https://github.com/xiuqhou/relation-detr/blob/HEAD/inference.py","link_basis":"first_harvest_node","language":"python","status":"ran_violates","verification_level":1,"contract_check":"VIOLATES","metamorphic_tier":"deterministic","behaviour_fingerprint":false,"licence":"Apache-2.0","inline_ok":true,"mcp_get_code":{"code_sha256":"fe838c2ebeafd791"}}]},"arxiv_metadata":null,"syntology_extracted_results":null}