AI Paper Atlas

Research map / Detection and segmentation / CNN features

Relation DETR: Exploring Explicit Position Relation Prior for Object Detection

Xiuquan Hou, Meiqin Liu, Senlin Zhang, Ping Wei, Badong Chen, Xuguang Lan

ECCV 2024 · Detection and segmentation › CNN features · approach: Efficient DETR query design · 90 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add position relation encoder to DINO (ablation)DINO ResNet-50 · COCO val2017 1xAP: 49.9 -> 50.3Improved
add VariFocal loss (ablation)DINO · COCO val2017 1xAP: 49.9 -> 50.9Improved
add position relation on top of VFL (ablation)DINO + VFL · COCO val2017 1xAP: 50.9 -> 51.2Improved
add contrast relation pipeline (ablation)DINO + VFL + Rel · COCO val2017 1xAP: 51.2 -> 51.7Improved
add relation encoder to Deformable-DETRDeformable-DETR · COCO val2017 1xAP: 45.4 -> 47.0Improved
add relation encoder to DAB-Deformable-DETRDAB-Deformable-DETR · COCO val2017 1xAP: 46.3 -> 48.3Improved
add relation encoder to DN-Deformable-DETRDN-Deformable-DETR · COCO val2017 1xAP: 47.2 -> 47.3No clear effect
apply hybrid matching to DINODINO · COCO val2017 1xAP: 49.9 -> 49.5Hurt
add relation encoder plus contrast pipeline to DINO instead of hybrid matchingDINO · COCO val2017 1xAP: 49.9 -> 51.2Improved
replace DINO with Relation-DETR on defect dataset MSSDDINO · MSSDAP@50:95: 51.0 -> 57.4Improved
replace DINO with Relation-DETR on CSDDINO · CSDAP@50:95: 53.0 -> 54.4Improved
use Relation-DETR vs DINO with VFLDINO with VFL · SA-Det-100kAP@50:95: 43.7 -> 45.0Improved

Later papers that compared with this method

6 later accepted papers report Relation-DETR as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy