Research map / Detection and segmentation / CNN features
Back-Tracking from Clarity: Self-Learning to See Text from Afar
Duc Tri Tran, Phi Le Nguyen, Minh Hoai Nguyen
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 carrier-based tracking to point-based tracking (ablation) | tracker · SceneText50 | AUC: 38.62 -> 44.71 | Improved |
| add carrier-based tracking to point-based tracking (ablation) | tracker · SceneText50 | Precision@20px: 57.44 -> 74.62 | Improved |
| add point-based tracking to carrier-based tracking (ablation) | tracker · SceneText50 | AUC: 34.68 -> 44.71 | Improved |
| use core polygon instead of convex hull (with RANSAC) (ablation) | tracker · SceneText50 | AUC: 37.40 -> 44.71 | Improved |
| use core polygon instead of convex hull (with RANSAC) (ablation) | tracker · SceneText50 | Precision@20px: 62.61 -> 74.62 | Improved |
| add RANSAC refinement to convex hull homography (ablation) | tracker · SceneText50 | AUC: 37.30 -> 37.40 | No clear effect |
| increase polygon vertices from 4 to 16-20 (ablation) | carrier tracker · SceneText50 | AUC: 32% -> 34.6% | Improved |
| increase polygon vertices beyond 16 (ablation) | carrier tracker · SceneText50 | AUC | No clear effect |
| train on tracked pseudo-labels instead of augmentation | Grounding DINO · SceneText50 | Precision: 14.21 -> 34.03 | Improved |
| train on tracked pseudo-labels instead of augmentation | Grounding DINO · SceneText50 | Recall: 29.77 -> 34.74 | Improved |
| train on tracked pseudo-labels instead of augmentation | DPText-DETR · SceneText50 | Recall: 21.80 -> 23.12 | Improved |
| train on tracked pseudo-labels instead of augmentation | DeepSolo · SceneText50 | Recall: 37.75 -> 40.27 | Improved |
Closest papers by meaning
- Beyond Detection: A Structure-Aware Framework for Scene Text Tracking (ICML 2026 · similarity 0.88)
- SemiETS: Integrating Spatial and Content Consistencies for Semi-Supervised End-to-end Text Spotting (CVPR 2025 · similarity 0.88)
- GoMatching: A Simple Baseline for Video Text Spotting via Long and Short Term Matching (NeurIPS 2024 · similarity 0.88)
- TIPS: Text-Image Pretraining with Spatial awareness (ICLR 2025 · similarity 0.86)
- OTE: Exploring Accurate Scene Text Recognition Using One Token (CVPR 2024 · similarity 0.85)
- 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.
