Research map / Vision-language and multimodal / Human and embodied modeling
RobustScanner: Dynamically Enhancing Positional Clues for Robust Text Recognition
Xiaoyu Yue, Zhanghui Kuang, Chenhao Lin, Hongbin Sun, Wayne Zhang
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 position enhancement branch (ablation) | RobustScanner · RandText | accuracy: 46.8 -> 83.4 | Improved |
| add hybrid branch (ablation) | RobustScanner · ICDAR 2015 | accuracy: 70.7 -> 77.1 | Improved |
| add hybrid branch (ablation) | RobustScanner · RandText | accuracy: 83.4 -> 81.2 | Hurt |
| add position aware module (ablation) | RobustScanner · RandText | accuracy: 68.0 -> 83.4 | Improved |
| add position aware module (ablation) | RobustScanner · ICDAR 2015 | accuracy: 61.8 -> 70.7 | Improved |
| add position aware module (ablation) | RobustScanner · SVTP | accuracy: 62.3 -> 74.4 | Improved |
| use position aware module instead of sine and cosine encoding (ablation) | RobustScanner · RandText | accuracy: 49.0 -> 83.4 | Improved |
| use position aware module instead of sine and cosine encoding (ablation) | RobustScanner · ICDAR 2015 | accuracy: 57.5 -> 70.7 | Improved |
| dynamic fusion instead of element-wise addition (ablation) | RobustScanner · ICDAR 2015 | accuracy: 75.5 -> 77.1 | Improved |
| dynamic fusion instead of concatenation (ablation) | RobustScanner · CUTE 80 | accuracy: 88.2 -> 90.3 | Improved |
| dynamic fusion instead of element-wise addition (ablation) | RobustScanner · RandText | accuracy: 82.4 -> 81.2 | Hurt |
| train additionally on real datasets (SAR vs RobustScanner comparison) | RobustScanner · SVTP | accuracy: 86.4 vs 82.9 | Hurt |
Later papers that compared with this method
5 later accepted papers report RobustScanner as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Image-to-Character-to-Word Transformers for Accurate Scene Text Recognition (TPAMI 2023 · better on 4 of 4 shared results)
- Toward Understanding WordArt: Corner-Guided Transformer for Scene Text Recognition (ECCV 2022 · better on 1 of 1 shared result)
- Multi-Granularity Prediction for Scene Text Recognition (ECCV 2022 · better on 1 of 1 shared result)
- ABINet++: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Spotting (TPAMI 2022 · better on 0 of 1 shared result)
- Read Like Humans: Autonomous, Bidirectional and Iterative Language Modeling for Scene Text Recognition (CVPR 2021 · better on 1 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OTE: Exploring Accurate Scene Text Recognition Using One Token (CVPR 2024 · similarity 0.89)
- Joint Visual Semantic Reasoning: Multi-Stage Decoder for Text Recognition (ICCV 2021 · similarity 0.89)
- Show, Attend and Read: A Simple and Strong Baseline for Irregular Text Recognition (AAAI 2019 · similarity 0.88)
- Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025 · similarity 0.88)
- OCRBench v2: An Improved Benchmark for Evaluating Large Multimodal Models on Visual Text Localization and Reasoning (NeurIPS 2025 · similarity 0.87)
- Boosting Semi-Supervised Scene Text Recognition via Viewing and Summarizing (NeurIPS 2024 · 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.
