Research map / Vision-language and multimodal / Cross-modal learning
UETrack: A Unified and Efficient Framework for Single Object Tracking
Ben Kang, Jie Zhao, Xin Chen, Wanting Geng, Bin Zhang, Lu Zhang, Dong Wang, Huchuan Lu
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 TP-MoE to the student (ablation) | UETrack-B · 5 benchmarks avg (Table 7) | AUC/F-score avg: -0.8% average when removed; LaSOT 67.9 -> 68.5 | Improved |
| use TP-MoE instead of gated MoE (ablation) | UETrack-B · Table 7 benchmarks + AGX speed | avg accuracy and FPS: avg -0.2% and 39 -> 60 FPS for gated | Improved |
| add local aggregation in TP-MoE (ablation) | UETrack-B · Table 7 benchmarks | avg accuracy: -0.3% average when removed | Improved |
| vary number of experts from 8 to 4/16/32 (ablation) | TP-MoE · Table 7 benchmarks | avg accuracy: -0.4%, -0.6%, -0.5% | Hurt |
| insert TP-MoE in more layers (last 2, last 3, even layers) (ablation) | UETrack-B · Table 7 benchmarks | avg accuracy: -0.2%, -0.6%, -0.6% | Hurt |
| add KL distillation on target distribution (ablation) | UETrack-B · Table 7 benchmarks | avg accuracy: +0.3% | Improved |
| add feature distillation (ablation) | UETrack-B · Table 7 benchmarks | avg accuracy: +0.3% -> +0.5% | Improved |
| add adaptive (sample-selective) distillation (ablation) | UETrack-B · Table 7 benchmarks | avg accuracy: +0.5% -> +1.0%; LaSOT 69.0 -> 69.2 | Improved |
| UETrack-B versus SUTrack-T on language tracking | UETrack-B · TNL2K | AUC: 60.9 (SUTrack-T) vs 58.0 | Hurt |
Closest papers by meaning
- SUTrack: Towards Simple and Unified Single Object Tracking (AAAI 2025 · similarity 0.89)
- UTPTrack: Towards Simple and Unified Token Pruning for Visual Tracking (CVPR 2026 · similarity 0.89)
- Single-Model and Any-Modality for Video Object Tracking (CVPR 2024 · similarity 0.88)
- OneTracker: Unifying Visual Object Tracking with Foundation Models and Efficient Tuning (CVPR 2024 · similarity 0.86)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.86)
- Collaborating Vision, Depth, and Thermal Signals for Multi-Modal Tracking: Dataset and Algorithm (NeurIPS 2025 · 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.
