Research map / Detection and segmentation / Open-set and few-shot detection
Detection Hub: Unifying Object Detection Datasets via Query Adaptation on Language Embedding
Lingchen Meng, Xiyang Dai, Yinpeng Chen, Pengchuan Zhang, Dongdong Chen, Mengchen Liu, Jianfeng Wang, Zuxuan Wu, Lu Yuan, Yu-Gang Jiang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| jointly train with dataset-aware queries and category-aligned language embeddings instead of training separately | Detection Hub · COCO | AP: 43.0 -> 45.3 | Improved |
| jointly train with dataset-aware queries and category-aligned language embeddings instead of training separately | Detection Hub · Object365 | AP: 22.2 -> 23.2 | Improved |
| jointly train with dataset-aware queries and category-aligned language embeddings instead of training separately | Detection Hub · Visual Genome | AP: 4.8 -> 5.7 | Improved |
| jointly train across the 11 UODB datasets using Detection Hub | Detection Hub · UODB | Avg: 64.2 -> 71.0 | Improved |
| use dataset-aware sampling rather than image-level instance sampling (ablation) | Detection Hub query adaptation · COCO, joint training | AP: 14.5 -> 45.3 | Improved |
| use dataset-aware sampling rather than global category sampling (ablation) | Detection Hub query adaptation · COCO, joint training | AP: 39.8 -> 45.3 | Improved |
| increase adapted query count from 100 to 300 (ablation) | Detection Hub · COCO, joint training | AP: 42.3 -> 45.3 | Improved |
| add category-aligned embedding (ablation) | Sparse-RCNN + Merge · COCO, separate training | AP: 40.6 -> 40.1 | Hurt |
| add category-aligned embedding (ablation) | Sparse-RCNN + Merge · COCO, joint training | AP: 40.4 -> 43.1 | Improved |
| add dataset-aware query adaptation to the decoder (ablation) | Sparse-RCNN + Merge with category-aligned embedding · COCO, joint training | AP: 43.1 -> 44.0 | Improved |
| add dataset-aware query adaptation to the RPN (ablation) | Detection Hub with decoder adaptation · COCO, joint training | AP: 44.0 -> 45.6 | Improved |
| add dataset-aware query adaptation to the RPN (ablation) | Detection Hub with decoder adaptation · Object365, joint training | AP: 21.9 -> 23.5 | Improved |
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.
- Dynamic-DETR — Dynamic DETR: End-to-End Object Detection With Dynamic Attention (ICCV 2021) · better on 2 of 2 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
- Simple Multi-Dataset Detection (CVPR 2022 · similarity 0.91)
- Cyclic Contrastive Knowledge Transfer for Open-Vocabulary Object Detection (ICLR 2025 · similarity 0.87)
- GCD: Advancing Vision-Language Models for Incremental Object Detection via Global Alignment and Correspondence Distillation (AAAI 2025 · similarity 0.87)
- ROD-MLLM: Towards More Reliable Object Detection in Multimodal Large Language Models (CVPR 2025 · similarity 0.86)
- Exploring Region-Word Alignment in Built-in Detector for Open-Vocabulary Object Detection (CVPR 2024 · similarity 0.86)
- DitHub: A Modular Framework for Incremental Open-Vocabulary Object Detection (NeurIPS 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.
