Research map / Detection and segmentation / CNN features
TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes
Shangbang Long, Jiaqiang Ruan, Wenjie Zhang, Xin He, Wenhao Wu, Cong Yao
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 |
|---|---|---|---|
| use the TextSnake disk/center-line representation instead of EAST's rotated-box regression | text detector · Total-Text | F-measure: 42.0 -> 78.4 | Improved |
| use the TextSnake representation instead of EAST | text detector · CTW1500 | F-measure: 60.4 -> 75.6 | Improved |
| use TextSnake instead of the CTD+TLOC curved-text baseline | text detector · CTW1500 | F-measure: 73.4 -> 75.6 | Improved |
| use TextSnake instead of PixelLink when trained only on ICDAR 2015 (no curved text) | text detector · Total-Text (cross-dataset) | F-measure: 53.1 -> 64.6 | Improved |
| use TextSnake instead of PixelLink when trained only on ICDAR 2015 (no curved text) | text detector · CTW1500 (cross-dataset) | F-measure: 46.4 -> 64.4 | Improved |
| use TextSnake single-scale instead of multi-scale WordSup on ICDAR 2015 | text detector · ICDAR 2015 | F-measure: 78.2 -> 82.6 | Improved |
| use TextSnake instead of PixelLink on ICDAR 2015 | text detector · ICDAR 2015 | F-measure: 83.7 -> 82.6 | Hurt |
| use TextSnake instead of PixelLink on MSRA-TD500 | text detector · MSRA-TD500 | F-measure: 77.8 -> 78.3 | Improved |
Later papers that compared with this method
6 later accepted papers report TextSnake as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.
- Arbitrary Shape Scene Text Detection With Adaptive Text Region Representation (CVPR 2019 · better on 2 of 2 shared results)
- Character Region Awareness for Text Detection (CVPR 2019 · better on 2 of 2 shared results)
- Look More Than Once: An Accurate Detector for Text of Arbitrary Shapes (CVPR 2019 · better on 2 of 2 shared results)
- Towards Robust Curve Text Detection With Conditional Spatial Expansion (CVPR 2019 · better on 2 of 2 shared results)
- TextDragon: An End-to-End Framework for Arbitrary Shaped Text Spotting (ICCV 2019 · better on 2 of 2 shared results)
- Learning Shape-Aware Embedding for Scene Text Detection (CVPR 2019 · 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
- EAST: An Efficient and Accurate Scene Text Detector (CVPR 2017 · similarity 0.89)
- Arbitrary Shape Scene Text Detection With Adaptive Text Region Representation (CVPR 2019 · similarity 0.88)
- Look More Than Once: An Accurate Detector for Text of Arbitrary Shapes (CVPR 2019 · similarity 0.88)
- CentripetalText: An Efficient Text Instance Representation for Scene Text Detection (NeurIPS 2021 · similarity 0.88)
- Character Region Awareness for Text Detection (CVPR 2019 · similarity 0.88)
- LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network (AAAI 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.
