Research map / Detection and segmentation / CNN features
Contextual Text Block Detection towards Scene Text Understanding
Chuhui Xue, Jiaxing Huang, Wenqing Zhang, Shijian Lu, Changhu Wang, Song Bai
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 |
|---|---|---|---|
| use CUTE-R50 instead of LINK-R50 | contextual text detection · ReCTS-Context (IoU=0.5) | Global Accuracy: 48.39 -> 51.55 | Improved |
| use CUTE-R50 instead of LINK-R50 | contextual text detection · SCUT-CTW-Context (IoU=0.5) | Global Accuracy: 18.88 -> 30.65 | Improved |
| use CUTE-R50 instead of LINK-R50 (Local Continuity) | contextual text detection · SCUT-CTW-Context (IoU=0.5) | Local Continuity: 3.33 -> 39.19 | Improved |
| use CUTE-R101 instead of CUTE-R50 backbone | CUTE · ReCTS-Context (IoU=0.5) | Global Accuracy: 51.55 -> 53.76 | Improved |
| use CUTE with GT integral boxes instead of LINK-R50 | grouping and ordering · SCUT-CTW-Context (GT integral boxes) | Global Accuracy: 22.84 -> 49.67 | Improved |
| add spatial embedding to feature embedding (ablation) | CUTE · SCUT-CTW-Context (ablation) | Global Accuracy: 1.94 -> 2.18 | No clear effect |
| add indexing embedding to feature embedding (ablation) | CUTE · SCUT-CTW-Context (ablation) | Global Accuracy: 1.94 -> 21.89 | Improved |
| add spatial embedding on top of feature + indexing embeddings (ablation) | CUTE · SCUT-CTW-Context (ablation) | Global Accuracy: 21.89 -> 49.67 | Improved |
| add CUTE to PSENet scene text detection | PSENet · SCUT-CTW scene text detection | mAP: 52.30 -> 53.69 | Improved |
| add CUTE to MSR scene text detection | MSR · SCUT-CTW scene text detection | mAP: 60.07 -> 61.80 | Improved |
| add CUTE to DETR scene text detection | DETR · SCUT-CTW scene text detection | mAP: 56.11 -> 57.37 | Improved |
| add CUTE to LINK scene text detection | LINK · SCUT-CTW scene text detection | mAP: 62.03 -> 62.84 | Improved |
Closest papers by meaning
- Character Region Awareness for Text Detection (CVPR 2019 · similarity 0.89)
- Revisiting Scene Text Recognition: A Data Perspective (ICCV 2023 · similarity 0.88)
- LayoutFormer: Hierarchical Text Detection Towards Scene Text Understanding (CVPR 2024 · similarity 0.87)
- OTE: Exploring Accurate Scene Text Recognition Using One Token (CVPR 2024 · similarity 0.87)
- Text Grouping Adapter: Adapting Pre-trained Text Detector for Layout Analysis (CVPR 2024 · similarity 0.87)
- Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 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.
