Research map / Detection and segmentation / CNN features
CF-DETR: Coarse-to-Fine Transformers for End-to-End 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 |
|---|---|---|---|
| Combine coarse global-context decoding with fine local-ROI refinement (ablation) | CF-DETR · COCO 2017 val | AP: 30.0 -> 46.5 | Helped |
| Add TEF to the CF-DETR baseline (ablation) | CF-DETR · COCO 2017 val | AP: 42.3 -> 43.8 | Helped |
| Add ASF to the CF-DETR baseline (ablation) | CF-DETR · COCO 2017 val | AP: 42.3 -> 44.2 | Helped |
| Add LCA to the CF-DETR baseline (ablation) | CF-DETR · COCO 2017 val | AP: 42.3 -> 43.4 | Helped |
| Add TEF, ASF, and LCA together (ablation) | CF-DETR · COCO 2017 val | AP: 42.3 -> 46.5 | Helped |
| Increase CF decoder layers from 1 to 2 (ablation) | CF-DETR · COCO 2017 val | AP: 35.8 -> 43.4 | Helped |
| Increase CF decoder layers from 6 to 12 (ablation) | CF-DETR · COCO 2017 val | AP: 46.5 -> 45.7 | Hurt |
| Use 300 object queries in training | CF-DETR-R50 · COCO 2017 val | AP: 46.5 -> 47.8 | Helped |
| Use test-time augmentations | CF-DETR with ResNeXt-101+DCN · COCO 2017 test-dev | AP: 50.7 -> 53.0 | Helped |
Closest papers by meaning
- Sparse DETR: Efficient End-to-End Object Detection with Learnable Sparsity (ICLR 2022 · similarity 0.92)
- Lite DETR: An Interleaved Multi-Scale Encoder for Efficient DETR (CVPR 2023 · similarity 0.90)
- DETRs Beat YOLOs on Real-time Object Detection (CVPR 2024 · similarity 0.90)
- MI-DETR: An Object Detection Model with Multi-time Inquiries Mechanism (CVPR 2025 · similarity 0.88)
- MS-DETR: Efficient DETR Training with Mixed Supervision (CVPR 2024 · similarity 0.87)
- CP-DETR: Concept Prompt Guide DETR Toward Stronger Universal Object Detection (AAAI 2025 · similarity 0.86)
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.