Research map / Architectures and efficiency / Attention and transformers
Adaptive and Background-Aware Vision Transformer for Real-Time UAV Tracking
Shuiwang Li, Yangxiang Yang, Dan Zeng, Xucheng Wang
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 background-aware weighting (wb=1.5) to the ponder loss of A-ViT (ablation) | A-ViT* tracker · DTB70 | precision: 84.1 -> 85.9 | Improved |
| use too-large background weight wb (e.g. 2.0) (ablation) | Aba-ViTrack · DTB70 | precision: 85.9 -> 82.5 | Hurt |
| tune ponder loss weight ap to 1.0e-4 versus other values (ablation) | Aba-ViTrack · DTB70 | precision: 82.9 -> 85.9 (max difference 3.0%) | Mixed |
| replace A-ViT with Aba-ViT backbone (ablation) | tracking framework · UAVDT | precision: 78.2 -> 83.4 | Improved |
| replace A-ViT with Aba-ViT backbone (ablation) | tracking framework · UAVTrack112_L | precision: 76.8 -> 81.1 | Improved |
| replace DeiT-tiny with Aba-ViT backbone (ablation) | tracking framework · UAV123 | precision: 82.8 -> 86.4 | Improved |
| replace DeiT-tiny with Aba-ViT backbone (ablation) | tracking framework · UAV123@10fps | AUC: 65.8 -> 65.5 | No clear effect |
| adaptive token halting (Aba-ViT) versus A-ViT in speed (ablation) | tracking framework · six UAV benchmarks | GPU FPS: 176.4 -> 181.5 | Improved |
| use efficient ViT unified framework instead of CNN-based trackers | UAV tracking · UAV123 | precision: 80.0 -> 86.4 | Improved |
Closest papers by meaning
- Learning Adaptive and View-Invariant Vision Transformer for Real-Time UAV Tracking (ICML 2024 · similarity 0.94)
- Similarity-Guided Layer-Adaptive Vision Transformer for UAV Tracking (CVPR 2025 · similarity 0.92)
- Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking (CVPR 2025 · similarity 0.89)
- LVPTrack: High Performance Domain Adaptive UAV Tracking with Label Aligned Visual Prompt Tuning (AAAI 2025 · similarity 0.86)
- MUST: The First Dataset and Unified Framework for Multispectral UAV Single Object Tracking (CVPR 2025 · similarity 0.85)
- Dynamic Semantic-Aware Correlation Modeling for UAV Tracking (NeurIPS 2025 · similarity 0.84)
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.
