Research map / Detection and segmentation / Open-set and few-shot detection
Towards Universal Object Detection by Domain Attention
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 data-driven domain attention and adapter branches to a shared detector | Universal object detection on UODB · UODB (11 datasets) | average mAP: 59.7 -> 63.8 | Helped |
| Replace per-dataset detector models with a shared domain-attentive detector | Universal object detection on UODB · UODB (11 datasets) | average mAP: 59.4 -> 63.8 | Helped |
| Use 6 adapters only in the first and middle blocks instead of 11 adapters (ablation) | universal+DA · UODB (11 datasets) | average mAP: 63.8 -> 63.9 | Helped |
| Fix assignments and average adapter responses instead of learning soft domain attention (ablation) | universal+DA · Five-dataset subset | average mAP: 75.1 -> 74.6 | Hurt |
| Use a single shared detector without domain attention | universal object detection · Five-dataset subset | average mAP: 74.2 -> 71.5 | Hurt |
| Increase the adapter count from 3 to 5 (ablation) | universal+DA · Five-dataset subset | average mAP: 74.5 -> 75.1 | Helped |
| Increase the adapter count from 5 to 7 (ablation) | universal+DA · Five-dataset subset | average mAP: 75.1 -> 75.6 | Helped |
| Increase the adapter count from 5 to 9 (ablation) | universal+DA · Five-dataset subset | average mAP: 75.1 -> 75.1 | No clear effect |
| Add domain-attention adapters | Faster R-CNN on DeepLesion · DeepLesion test | Sensitivity at 4 FPs per image: 82.44 -> 87.29 | Helped |
| Add domain-attention adapters | Faster R-CNN on Clipart · Clipart test | mAP: 32.1 -> 54.6 | Helped |
| Add domain-attention adapters | Faster R-CNN on VOC · VOC 2007 test | mAP: 78.5 -> 79.6 | Helped |
| Add domain-attention adapters | Faster R-CNN on WiderFace · WiderFace Val | Hard score: 0.556 -> 0.587 | Helped |
Closest papers by meaning
- SEEN-DA: SEmantic ENtropy guided Domain-aware Attention for Domain Adaptive Object Detection (CVPR 2025 · similarity 0.89)
- Improving Single Domain-Generalized Object Detection: A Focus on Diversification and Alignment (CVPR 2024 · similarity 0.88)
- Towards Single-Source Domain Generalized Object Detection via Causal Visual Prompts (NeurIPS 2025 · similarity 0.88)
- CP-DETR: Concept Prompt Guide DETR Toward Stronger Universal Object Detection (AAAI 2025 · similarity 0.87)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.87)
- Unified Domain Generalization and Adaptation for Multi-View 3D Object Detection (NeurIPS 2024 · 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.