Research map / Detection and segmentation / Open-set and few-shot detection
Context-Transformer: Tackling Object Confusion for Few-Shot 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 |
|---|---|---|---|
| add Context-Transformer and preserved source OBJ | SSD fine-tuning baseline · VOC07 test 5-shot (COCO-60 source) | mAP: 39.4 -> 43.8 | Helped |
| add Context-Transformer between source and target OBJ (ablation) | source-OBJ transfer · VOC07 test 5-shot (COCO-60 source) | mAP: 40.9 -> 43.8 | Helped |
| preserve source OBJ under target OBJ (ablation) | Context-Transformer · VOC07 test 5-shot (COCO-60 source) | mAP: 41.5 -> 43.8 | Helped |
| unload Context-Transformer at test time (ablation) | Context-Transformer · VOC07 test 5-shot (COCO-60 source) | mAP: 43.8 -> 43.4 | No clear effect |
| max pooling to build contextual fields (ablation) | Context-Transformer · VOC07 test 5-shot (COCO-60 source) | mAP: 42.5 -> 43.8 | Helped |
| add residual FC embedding layers (ablation) | Context-Transformer · VOC07 test 5-shot (COCO-60 source) | mAP: 42.2 -> 43.8 | Helped |
| cosine instead of Euclidean affinity (ablation) | Context-Transformer · VOC07 test 5-shot (COCO-60 source) | mAP: 43.5 -> 43.8 | No clear effect |
| share target OBJ across scales (ablation) | Context-Transformer · VOC07 test 5-shot (COCO-60 source) | mAP: 41.4 -> 43.8 | Helped |
| Context-Transformer vs Non-local attention | transfer framework · VOC07 test 5-shot (COCO-60 source) | mAP: 41.0 -> 43.8 | Helped |
| Context-Transformer with more shots | SSD baseline · VOC07 test | mAP: 1-shot +5.5 (21.5 -> 27.0); all +0.8 (80.7 -> 81.5) | Helped |
| Context-Transformer in incremental few-shot | incremental detection · VOC split1 1-shot | novel mAP: 23.9 -> 34.2 | Helped |
Closest papers by meaning
- Temporal Object-Aware Vision Transformer for Few-Shot Video Object Detection (AAAI 2026 · similarity 0.89)
- LSTD: A Low-Shot Transfer Detector for Object Detection (AAAI 2018 · similarity 0.89)
- Context Enhanced Transformer for Single Image Object Detection in Video Data (AAAI 2024 · similarity 0.88)
- Multi-Perspective Data Augmentation for Few-shot Object Detection (ICLR 2025 · similarity 0.87)
- Few-Shot Object Detection with Foundation Models (CVPR 2024 · similarity 0.87)
- Revisiting Few-Shot Object Detection with Vision-Language Models (NeurIPS 2024 · similarity 0.87)
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.