Research map / Detection and segmentation / Tracking
Foreground-Background Distribution Modeling Transformer for Visual Object Tracking
Dawei Yang, Jianfeng He, Yinchao Ma, Qianjin Yu, Tianzhu Zhang
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 FBAL module (pseudo bbox loss) to plain attention (ablation) | OSTrack-style baseline · LaSOT | AUC: 68.5 -> 68.8 | Improved |
| replace plain attention with distribution-aware attention (DA2) (ablation) | OSTrack-style baseline · LaSOT | AUC: 68.5 -> 69.4 | Improved |
| add FBAL on top of DA2 (ablation) | F-BDMTrack-256 · LaSOT | AUC: 69.4 -> 69.9 | Improved |
| increase number of FB-agents from 4 to 16 (ablation) | F-BDMTrack-256 · LaSOT | AUC: 69.9 -> 69.0 | Hurt |
| reduce number of FB-agents from 4 to 1 (ablation) | F-BDMTrack-256 · LaSOT | AUC: 69.9 -> 69.3 | Hurt |
| apply pseudo bbox generation in blocks 6-12 instead of none (ablation) | F-BDMTrack-256 · LaSOT | AUC: 69.0 -> 69.9 | Improved |
| apply pseudo bbox generation in all blocks 1-12 instead of 6-12 (ablation) | F-BDMTrack-256 · LaSOT | AUC: 69.9 -> 69.2 | Hurt |
| use F-BDMTrack-256 instead of OSTrack-256 | OSTrack-256 · GOT-10k | AO: 71.0 -> 72.7 | Improved |
| use F-BDMTrack-256 instead of OSTrack-256 | OSTrack-256 · TNL2K | AUC: 54.3 -> 56.4 | Improved |
| use F-BDMTrack-256 instead of OSTrack-256 | OSTrack-256 · LaSOText | AUC: 47.4 -> 47.9 | Improved |
Closest papers by meaning
- Transformer Meets Tracker: Exploiting Temporal Context for Robust Visual Tracking (CVPR 2021 · similarity 0.87)
- Backbone Is All Your Need: A Simplified Architecture for Visual Object Tracking (ECCV 2022 · similarity 0.87)
- Compact Transformer Tracker with Correlative Masked Modeling (AAAI 2023 · similarity 0.87)
- Correlation-Embedded Transformer Tracking: A Single-Branch Framework (TPAMI 2024 · similarity 0.87)
- Target Focused Shallow Transformer Framework for Efficient Visual Tracking (AAAI 2024 · similarity 0.84)
- VastTrack: Vast Category Visual Object Tracking (NeurIPS 2024 · similarity 0.83)
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.
