Research map / Architectures and efficiency / Attention and transformers
Compact Transformer Tracker with Correlative Masked Modeling
Read the paperCheck 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 all three correlative reconstruction streams (s2s, t2t, and s2t) (ablation) | CTTrack-B · LaSOT | AUC: 64.0 -> 65.8 | Helped |
| Add search-token self-reconstruction (s2s) (ablation) | CTTrack-B · LaSOT | AUC: 64.0 -> 64.7 | Helped |
| Add template-token self-reconstruction (t2t) (ablation) | CTTrack-B · LaSOT | AUC: 64.0 -> 64.4 | Helped |
| Add search-to-template reconstruction (s2t) (ablation) | CTTrack-B · LaSOT | AUC: 64.0 -> 64.4 | Helped |
| Use a 75% masking ratio rather than a 25% ratio (ablation) | CTTrack-B · LaSOT | AUC: 64.6 -> 65.8 | Helped |
| Increase masking ratio from 75% to 90% (ablation) | CTTrack-B · LaSOT | AUC: 65.8 -> 64.9 | Hurt |
| Add fixed-interval online template updating without confidence filtering (ablation) | CTTrack-B · LaSOT | AUC: 65.8 -> 64.9 | Hurt |
| Add confidence-based score-head filtering to online template updating (ablation) | CTTrack-B with fixed-interval updating · LaSOT | AUC: 64.9 -> 66.0 | Helped |
| Remove template cross-information stream 2 from packed self-attention (ablation) | CTTrack · LaSOT | AUC: 61.7 -> 64.0 | Helped |
| Remove template self-information stream 1 from packed self-attention (ablation) | CTTrack · LaSOT | AUC: 61.7 -> 60.6 | Hurt |
| Remove search self-information stream 4 from packed self-attention (ablation) | CTTrack · LaSOT | AUC: 61.7 -> 58.8 | Hurt |
| Use ViT-Large instead of ViT-Base (ablation) | CTTrack · LaSOT | AUC: 67.8 -> 69.8 | Helped |
Closest papers by meaning
- Target Focused Shallow Transformer Framework for Efficient Visual Tracking (AAAI 2024 · similarity 0.89)
- MIMTrack: In-Context Tracking via Masked Image Modeling (AAAI 2025 · similarity 0.87)
- Sequential Fusion Based Multi-Granularity Consistency for Space-Time Transformer Tracking (AAAI 2024 · similarity 0.86)
- Cross-modulated Attention Transformer for RGBT Tracking (AAAI 2025 · similarity 0.86)
- Video Diffusion Models Excel at Tracking Similar-Looking Objects Without Supervision (NeurIPS 2025 · similarity 0.86)
- OVTR: End-to-End Open-Vocabulary Multiple Object Tracking with Transformer (ICLR 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.