Research map / Architectures and efficiency / Attention and transformers
High-Performance Transformer Tracking
Xin Chen, Bin Yan, Jiawen Zhu, Huchuan Lu, Xiang Ruan, Dong 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 postprocessing window penalty (ablation) | TransT · LaSOT | AUC: 62.9 -> 64.9 | Improved |
| use proposed fusion network instead of original transformer (ablation) | TransT · LaSOT | AUC: 62.3 -> 64.9 | Improved |
| feed template to encoder-side as in proposed order instead of swapping inputs (ablation) | TransT · TrackingNet | AUC: 77.8 -> 81.4 | Improved |
| add ego-context augment (ECA) (ablation) | TransT · LaSOT | AUC: 62.9 -> 64.9 | Improved |
| use cross-feature augment instead of correlation (ablation) | TransT · LaSOT | AUC: 57.7 -> 64.9 | Improved |
| remove both attention modules (correlation only) (ablation) | TransT · LaSOT | AUC: 64.9 -> 47.7 | Hurt |
| use N=4 fusion layers instead of N=2 (ablation) | TransT · LaSOT | AUC: 64.2 -> 64.9 | Improved |
| increase fusion layers from N=4 to N=6 (ablation) | TransT · LaSOT | AUC: 64.9 -> 64.4 | Hurt |
| use ResNet101 instead of ResNet50 backbone (ablation) | TransT · LaSOT | AUC: 64.9 -> 64.2 | Hurt |
| use ResNet50 instead of ResNet18 backbone (ablation) | TransT · LaSOT | AUC: 63.9 -> 64.9 | Improved |
| add segmentation branch to TransT | TransT · VOT2021 | EAO: 30.0 -> 50.7 | Improved |
| use SEG segmentation branch instead of SEG-na (ablation) | TransT-M · VOT2021 | EAO: 53.7 -> 55.0 | Improved |
Closest papers by meaning
- Transformer Tracking (CVPR 2021 · similarity 0.97)
- AiATrack: Attention in Attention for Transformer Visual Tracking (ECCV 2022 · similarity 0.91)
- Compact Transformer Tracker with Correlative Masked Modeling (AAAI 2023 · similarity 0.90)
- Target Focused Shallow Transformer Framework for Efficient Visual Tracking (AAAI 2024 · similarity 0.88)
- Cross-modulated Attention Transformer for RGBT Tracking (AAAI 2025 · similarity 0.87)
- Sequential Fusion Based Multi-Granularity Consistency for Space-Time Transformer Tracking (AAAI 2024 · similarity 0.85)
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.
