Research map / Detection and segmentation / Tracking
Backbone Is All Your Need: A Simplified Architecture for Visual Object Tracking
Boyu Chen, Peixia Li, Lei Bai, Lei Qiao, Qiuhong Shen, Bo Li, Weihao Gan, Wei Wu, Wanli Ouyang
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 |
|---|---|---|---|
| use SimTrack one-stream framework instead of STARK-SV with same backbone (ablation) | ViT-B/32 · LaSOT | AUC: 62.5 -> 66.2 | Improved |
| use SimTrack framework instead of STARK-SV (ablation) | ViT-B/16 · LaSOT | AUC: 66.8 -> 69.3 | Improved |
| use SimTrack framework instead of STARK-SV (ablation) | ViT-L/14 · LaSOT | AUC: 69.2 -> 70.5 | Improved |
| add +Sim framework (no position embedding or foveal window) (ablation) | STARK-SV ViT-B/16 · TNL2K | AUC: 52.2 -> 53.5 | Improved |
| add distinguishable position embedding (ablation) | SimTrack · TNL2K | AUC: 53.5 -> 54.0 | Improved |
| add foveal window strategy (ablation) | SimTrack · TNL2K | AUC: 54.0 -> 54.8 | Improved |
| add transformer decoder layers after backbone (0 to 6) (ablation) | SimTrack · TNL2K | AUC: 54.8 -> 54.7 | No clear effect |
| reduce interaction to 50% of blocks (ablation) | SimTrack · TNL2K | AUC: 54.8 -> 52.3 | Hurt |
| reduce interaction to 25% of blocks (ablation) | SimTrack · TNL2K | AUC: 52.3 -> 49.8 | Hurt |
| use MAE pre-training instead of DeiT (ablation) | SimTrack ViT-B/16 · LaSOT | AUC: 66.9 -> 70.3 | Improved |
| use CLIP pre-training instead of MoCo-v3 (ablation) | SimTrack ViT-B/16 · LaSOT | AUC: 66.4 -> 69.3 | Improved |
| remove a(e,s) and sW_V terms for Swin adaptation (ablation) | SimTrack ViT · LaSOT | AUC: 69.3 -> 69.1 | No clear effect |
Closest papers by meaning
- Correlation-Embedded Transformer Tracking: A Single-Branch Framework (TPAMI 2024 · similarity 0.89)
- Transformer Meets Tracker: Exploiting Temporal Context for Robust Visual Tracking (CVPR 2021 · similarity 0.88)
- Compact Transformer Tracker with Correlative Masked Modeling (AAAI 2023 · similarity 0.88)
- OmniTracker: Unifying Visual Object Tracking by Tracking-With-Detection (TPAMI 2025 · similarity 0.88)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.86)
- UTPTrack: Towards Simple and Unified Token Pruning for Visual Tracking (CVPR 2026 · 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.
