Research map / Detection and segmentation / Open-set and few-shot detection
AirDet: Few-Shot Detection without Fine-Tuning for Autonomous Exploration
Bowen Li, Chen Wang, Pranay Reddy, Seungchan Kim, Sebastian Scherer
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 |
|---|---|---|---|
| add PRE module to baseline (ablation) | A-RPN baseline · COCO 3-shot (no fine-tuning) | AP: 4.80 -> 5.15 | Improved |
| add GLR on top of PRE (ablation) | AirDet · COCO 3-shot (no fine-tuning) | AP: 5.15 -> 5.59 | Improved |
| add SCS on top of PRE and GLR (ablation) | AirDet · COCO 3-shot (no fine-tuning) | AP: 5.59 -> 6.50 | Improved |
| add PRE to baseline (ablation) | A-RPN baseline · COCO 5-shot (no fine-tuning) | AP75: 5.53 -> 5.34 | Hurt |
| add all three modules (PRE, GLR, SCS) (ablation) | A-RPN baseline · COCO 5-shot (no fine-tuning) | AP: 5.73 -> 7.27 | Improved |
| use ResNet101 instead of ResNet50 backbone (ablation) | AirDet · COCO 1-shot | AP: 4.64 -> 6.00 | Improved |
| use ResNet101 instead of ResNet50 backbone (ablation) | AirDet · COCO 5-shot | AP: 7.43 -> 7.63 | No clear effect |
| fine-tune AirDet on novel shots | AirDet · COCO 2-shot | AP: 6.58 -> 8.73 | Improved |
| fine-tune AirDet on novel shots | AirDet · COCO 10-shot | AP: 8.7 -> 13.0 | Improved |
| pre-train with 60 instead of 50 base classes (ablation) | AirDet · VOC-2012 3-shot cross-domain | AP: 6.54 -> 16.89 | Improved |
| pre-train with 60 instead of 55 base classes (ablation) | AirDet · VOC-2012 5-shot cross-domain | AP: 11.76 -> 17.83 | Improved |
| use same rather than different support objects at test | AirDet · real-world SubT test | AP: 10.4 -> 16.4 | Improved |
Closest papers by meaning
- PromptDet: A Lightweight 3D Object Detection Framework with LiDAR Prompts (AAAI 2025 · similarity 0.86)
- A High-Resolution Dataset for Instance Detection with Multi-View Object Capture (NeurIPS 2023 · similarity 0.85)
- MobileDets: Searching for Object Detection Architectures for Mobile Accelerators (CVPR 2021 · similarity 0.85)
- RemDet: Rethinking Efficient Model Design for UAV Object Detection (AAAI 2025 · similarity 0.84)
- Revisiting Few-Shot Object Detection with Vision-Language Models (NeurIPS 2024 · similarity 0.83)
- SAFDNet: A Simple and Effective Network for Fully Sparse 3D Object Detection (CVPR 2024 · similarity 0.83)
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.
