Research map / Label-efficient and robust learning / Knowledge distillation
Dual-branch Distilled Transformer for Efficient Asymmetric UAV Tracking
Hongtao Yang, Bineng Zhong, Qihua Liang, Yaozong Zheng, Xiantao Hu, Yuanliang Xue, Shuxiang Song
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 feature-level distillation to the student (ablation) | EATrack-DeiT · UAV123 | success: 66.1 -> 67.0 | Improved |
| add prediction-level distillation on top of feature-level distillation (ablation) | EATrack-DeiT · UAV123 | precision and success: 87.6 -> 89.0 precision, 67.0 -> 68.1 success | Improved |
| add temporal cues at inference (ablation) | EATrack-DeiT · DTB70 | success: 63.3 -> 66.5 | Improved |
| add temporal cues at inference (ablation) | EATrack-DeiT · UAVDT | success: 56.6 -> 60.2 | Improved |
| add temporal cues at inference (ablation) | EATrack-DeiT · VisDrone | success: 59.8 -> 65.6 | Improved |
| add temporal cues at inference (ablation) | EATrack-DeiT · UAV123 | success: 68.1 -> 68.1 | No clear effect |
| increase Stage 1 depth from 4 to 5 layers (ablation) | EATrack-DeiT · UAV123 | success: 66.4 -> 67.4 | Improved |
| increase Stage 1 depth from 5 to 6 layers (ablation) | EATrack-DeiT · UAV123 | success: 67.4 -> 68.1 | Improved |
| increase Stage 2 depth from 1 to 2 layers (ablation) | EATrack-DeiT · UAV123 | success: 67.0 -> 68.1 | Improved |
| increase Stage 2 depth from 1 to 2 layers (speed cost) (ablation) | EATrack-DeiT · UAV123 | speed (fps): 263.4 -> 241.9 | Hurt |
| EATrack-DeiT vs SGLATrack and ORTrack inference speed on GPU | EATrack · UAV123 speed | FPS: 222.8 -> 241.9 vs SGLATrack; 211.8 for ORTrack | Improved |
Closest papers by meaning
- Learning Occlusion-Robust Vision Transformers for Real-Time UAV Tracking (CVPR 2025 · similarity 0.87)
- LVPTrack: High Performance Domain Adaptive UAV Tracking with Label Aligned Visual Prompt Tuning (AAAI 2025 · similarity 0.86)
- Learning Adaptive and View-Invariant Vision Transformer for Real-Time UAV Tracking (ICML 2024 · similarity 0.86)
- Similarity-Guided Layer-Adaptive Vision Transformer for UAV Tracking (CVPR 2025 · similarity 0.84)
- Dynamic Semantic-Aware Correlation Modeling for UAV Tracking (NeurIPS 2025 · similarity 0.83)
- Event Stream-based Visual Object Tracking: A High-Resolution Benchmark Dataset and A Novel Baseline (CVPR 2024 · similarity 0.82)
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.
