Research map / Detection and segmentation / CNN features
Character Region Attention For Text Spotting
Youngmin Baek, Seung Shin, Jeonghun Baek, Sungrae Park, Junyeop Lee, Daehyun Nam, Hwalsuk Lee
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 Character Region Attention (ablation) | CRAFTS · IC13 end-to-end | H-mean: 91.9 -> 92.2 (+0.3) | Improved |
| Add Character Region Attention (ablation) | CRAFTS · IC15 end-to-end | H-mean: 73.2 -> 74.9 (+1.7) | Improved |
| Add Character Region Attention (ablation) | CRAFTS · TotalText end-to-end | H-mean: 77.5 -> 78.7 (+1.2) | Improved |
| Propagate recognition loss to detection stage (ablation) | CRAFTS · ICDAR2015 | character error length: 1,251 -> 1,147 (-8.3%) | Improved |
| Propagate recognition loss to detection stage (ablation) | CRAFTS · TotalText | character error length: 5,050 -> 4,521 (-10.4%) | Improved |
| Add orientation estimation (ablation) | CRAFTS · IC15 end-to-end | H-mean: 74.1 -> 74.9 | Improved |
| Add orientation estimation (ablation) | CRAFTS · TotalText end-to-end | H-mean: 77.5 -> 78.7 | Improved |
| Replace VGG-16 CRAFT detector with ResNet50-based CRAFTS | CRAFT detector · inference speed | FPS: 8.6 FPS vs 8.3 (1280 side) | Improved |
| Reduce input longer side from 2560 to 960 | CRAFTS · inference speed | FPS: 5.4 -> 9.9 | Improved |
| Pretrain recognizer with other multi-language datasets with detector frozen | CRAFTS · IC19-MLT | end-to-end H-mean | Improved |
| Use CRAFTS vs MaskTextSpotter on generic E2E | CRAFTS · IC13 end-to-end generic | H-mean: 88.2 -> 92.2 | Improved |
| Use CRAFTS vs CharNet on IC15 detection | CRAFTS · IC15 detection | H-mean: 91.5 -> 87.1 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FOTS — FOTS: Fast Oriented Text Spotting With a Unified Network (CVPR 2018) · better on 1 of 1 shared result
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
- Towards End-to-End Text Spotting in Natural Scenes (TPAMI 2021 · similarity 0.90)
- MANGO: A Mask Attention Guided One-Stage Scene Text Spotter (AAAI 2021 · similarity 0.89)
- Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025 · similarity 0.88)
- Towards Unified Scene Text Spotting Based on Sequence Generation (CVPR 2023 · similarity 0.88)
- SPTS v2: Single-Point Scene Text Spotting (TPAMI 2023 · similarity 0.87)
- SwinTextSpotter v2: Towards Better Synergy for Scene Text Spotting (IJCV 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.
