Research map / Detection and segmentation / CNN features
Rank-DETR for High Quality Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| add rank-adaptive classification head (RCH) (ablation) | H-DETR · COCO val2017 R50 12ep, H-DETR base | AP: 48.7 -> 48.9 | Helped |
| add query ranking layer (QRL) (ablation) | H-DETR · COCO val2017 R50 12ep, H-DETR base | AP: 48.7 -> 49.0 | Helped |
| add GIoU-aware classification loss (GCL) (ablation) | H-DETR · COCO val2017 R50 12ep, H-DETR base | AP: 48.7 -> 49.4 | Helped |
| add high-order matching cost (HMC) (ablation) | H-DETR · COCO val2017 R50 12ep, H-DETR base | AP: 48.7 -> 49.3 | Helped |
| add all four components (RCH, QRL, GCL, HMC) (ablation) | H-DETR · COCO val2017 R50 12ep, H-DETR base | AP: 48.7 -> 50.2 | Helped |
| add QRL to the complete method (ablation) | Rank-DETR · COCO val2017 R50 12ep, H-DETR base | AP: 49.5 -> 50.2 | Helped |
| use (GIoU+1)/2 as classification target instead of IoU (ablation) | GCL · COCO val2017 R50 12ep, H-DETR base | AP: 50.0 -> 50.2 | Helped |
| raise matching cost IoU exponent from 1 to 4 (ablation) | HMC · COCO val2017 R50 12ep, H-DETR base | AP: 48.3 -> 50.2 | Helped |
| raise matching cost IoU exponent from 4 to 6 (ablation) | HMC · COCO val2017 R50 12ep, H-DETR base | AP: 50.2 -> 50.0 | No clear effect |
| use GCL instead of varifocal loss (ablation) | classification loss · COCO val2017 R50 12ep, H-DETR base | AP: 49.5 -> 49.8 | Helped |
| apply Rank-DETR to H-Deformable-DETR (R50, 12 epochs) | H-Deformable-DETR · COCO val2017 | AP: 48.7 -> 50.2 | Helped |
| apply Rank-DETR to DINO-DETR (Swin-L, 12 epochs) | DINO-DETR · COCO val2017 | AP: 56.8 -> 57.6 | Helped |
Closest papers by meaning
- RankDetNet: Delving Into Ranking Constraints for Object Detection (CVPR 2021 · similarity 0.90)
- DETRs Beat YOLOs on Real-time Object Detection (CVPR 2024 · similarity 0.87)
- MS-DETR: Efficient DETR Training with Mixed Supervision (CVPR 2024 · similarity 0.87)
- V-DETR: DETR with Vertex Relative Position Encoding for 3D Object Detection (ICLR 2024 · similarity 0.86)
- MI-DETR: An Object Detection Model with Multi-time Inquiries Mechanism (CVPR 2025 · similarity 0.86)
- CP-DETR: Concept Prompt Guide DETR Toward Stronger Universal Object Detection (AAAI 2025 · similarity 0.85)
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.