Research map / Detection and segmentation / Tracking
Transformer Meets Tracker: Exploiting Temporal Context for Robust Visual Tracking
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the full temporal transformer (ablation) | Siamese tracker · GOT-10k test | AO: 62.0 -> 67.3; 5.3%↑ | Helped |
| add the full temporal transformer (ablation) | DiMP tracker · GOT-10k test | AO: 66.7 -> 68.8; 2.1%↑ | Helped |
| add template encoder without decoder (ablation) | Siamese tracker · GOT-10k test | AO: 62.0 -> 63.8 | Helped |
| add template encoder without decoder (ablation) | DiMP tracker · GOT-10k test | AO: 66.7 -> 67.3 | Helped |
| add feature propagation (ablation) | Siamese tracker · GOT-10k test | AO: 62.0 -> 66.3; 4.3%↑ | Helped |
| add feature propagation (ablation) | DiMP tracker · GOT-10k test | AO: 66.7 -> 68.1; 1.4%↑ | Helped |
| add mask propagation (ablation) | Siamese tracker · GOT-10k test | AO: 62.0 -> 67.1; 5.1%↑ | Helped |
| add mask propagation (ablation) | DiMP tracker · GOT-10k test | AO: 66.7 -> 67.8; 1.1%↑ | Helped |
| remove self-attention weight-sharing (ablation) | TrSiam · GOT-10k test | AO (%): 67.3 -> 63.4 | Hurt |
| add feed-forward layers (ablation) | TrSiam · GOT-10k test | AO (%): 67.3 -> 67.0 | Hurt |
| add feed-forward layers (ablation) | TrSiam · GOT-10k test | Speed (FPS): 35 -> 22 | Hurt |
| increase attention heads from 1 to 4 (ablation) | TrSiam · GOT-10k test | AO (%): 67.3 -> 67.6 | Helped |
Closest papers by meaning
- Track-On: Transformer-based Online Point Tracking with Memory (ICLR 2025 · similarity 0.89)
- Sequential Fusion Based Multi-Granularity Consistency for Space-Time Transformer Tracking (AAAI 2024 · similarity 0.88)
- UTPTrack: Towards Simple and Unified Token Pruning for Visual Tracking (CVPR 2026 · similarity 0.88)
- MIMTrack: In-Context Tracking via Masked Image Modeling (AAAI 2025 · similarity 0.87)
- Efficient Motion Prompt Learning for Robust Visual Tracking (ICML 2025 · similarity 0.87)
- Emergent Temporal Correspondences from Video Diffusion Transformers (NeurIPS 2025 · similarity 0.87)
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.