Research map / Detection and segmentation / CNN features
QueryDet: Cascaded Sparse Query for Accelerating High-Resolution Small Object Detection
Check 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 high-resolution P2 without loss re-balancing (ablation) | RetinaNet · COCO mini-val | AP: 37.46 -> 36.10 | Hurt |
| Re-balance FPN losses when adding P2 (ablation) | RetinaNet with high-resolution P2 · COCO mini-val | AP: 36.10 -> 38.11 | Helped |
| Add loss re-balancing to the original baseline (ablation) | RetinaNet · COCO mini-val | AP: 37.46 -> 37.66 | Helped |
| Add Query Head objectiveness supervision (ablation) | High-resolution, loss-rebalanced RetinaNet · COCO mini-val | AP: 38.11 -> 38.53 | Helped |
| Add Query Head objectiveness supervision (ablation) | High-resolution, loss-rebalanced RetinaNet · COCO mini-val | AP_S: +1.58 AP_S | Helped |
| Add CSQ to sparse-query detection (ablation) | QueryDet · COCO mini-val | FPS: 4.85 -> 14.88 | Helped |
| Add CSQ (ablation) | QueryDet · COCO mini-val | AP: 38.53 -> 38.36 | No clear effect |
| Add high-resolution features and QueryDet | RetinaNet · VisDrone validation | AP: 26.21 -> 28.35 | Helped |
| Add CSQ | QueryDet · VisDrone validation | FPS: 1.16 -> 2.75 | Helped |
| Start CSQ at P4 rather than P6 (ablation) | QueryDet · COCO mini-val | AP: 37.91 -> 38.36 | Helped |
| Start CSQ at P4 rather than P6 (ablation) | QueryDet · COCO mini-val | FPS: 13.42 -> 14.88 | Helped |
| Replace CSQ with Crop Query (ablation) | QueryDet · COCO mini-val | FPS: 14.88 -> 10.49 | Hurt |
Closest papers by meaning
- AdaMixer: A Fast-Converging Query-Based Object Detector (CVPR 2022 · similarity 0.88)
- Search and Detect: Training-Free Long Tail Object Detection via Web-Image Retrieval (CVPR 2025 · similarity 0.88)
- MegDet: A Large Mini-Batch Object Detector (CVPR 2018 · similarity 0.87)
- iDETR: Implicit DETR for Tiny Object Detection (NeurIPS 2026 · similarity 0.87)
- Query-Based Hard-Image Retrieval for Object Detection at Test Time (AAAI 2023 · similarity 0.86)
- CQ-DINO: Mitigating Gradient Dilution via Category Queries for Vast Vocabulary Object Detection (NeurIPS 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.