Research map / Detection and segmentation / Tracking
Beyond Detection: A Structure-Aware Framework for Scene Text Tracking
Chenmin Yu, Liu Yu, Daiqing Wu, Gengluo Li, Zeyu Chen, Yu ZHOU
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 Predictive Token Rectification (PTR) (ablation) | SymTrack baseline tracker · ArTVideoSOT | AUC (%): 69.50 -> 74.46 (+4.96) | Improved |
| add Cross-Expert Calibration (CEC) on top of PTR (ablation) | SymTrack · ArTVideoSOT | AUC (%): 74.46 -> 76.58 (+2.12) | Improved |
| add training-free Adaptive Inference Engine (AIE) to CEC+PTR model (ablation) | SymTrack · ArTVideoSOT | AUC (%): 76.58 -> 77.74 (+1.16) | Improved |
| add AIE alone to the baseline without PTR/CEC (ablation) | baseline tracker · ArTVideoSOT | AUC (%): 69.50 -> 70.85 | Improved |
| add AIE (adaptive search region and temporal regularization) (ablation) | SymTrack · DSTextSOT | Search Region Coverage / Avg. IoU: SRC 83.27 -> 95.25; IoU 69.54 -> 71.96 | Improved |
| remove the TokenFD textual expert (text-specific branch) (ablation) | SymTrack · ArTVideoSOT | AUC (%) / speed: AUC 77.74 -> 75.45; fps 22.07 -> 89.02; params 395.917M -> 92.674M | Mixed |
| fine-tune the strongest generic tracker ODTrack on the STT training splits | ODTrack · BOVTextSOT | AUC (%): 64.74 -> 67.23 | Improved |
| use detection-free tracking instead of detection-based VTS pipeline | SymTrack vs TransDETR · DSTextSOT | AUC (%): 5.08 -> 70.66 | Improved |
| CEC removed: response map without text-specific guidance (ablation) | SymTrack · Qualitative attention/score maps | localization confidence | Hurt |
Closest papers by meaning
- Back-Tracking from Clarity: Self-Learning to See Text from Afar (ECCV 2026 · similarity 0.88)
- GoMatching: A Simple Baseline for Video Text Spotting via Long and Short Term Matching (NeurIPS 2024 · similarity 0.87)
- TextDS: Parameter-Efficient Representation Alignment for Scene Text Detection under Distribution Shifts (ECCV 2026 · similarity 0.87)
- Leveraging Text-to-Image Diffusion Models for Unsupervised Visual Object Tracking (TPAMI 2026 · similarity 0.87)
- End-to-End Video Text Spotting with Transformer (IJCV 2024 · similarity 0.87)
- Dense Video Object Captioning from Disjoint Supervision (ICLR 2025 · 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.
