Research map / Detection and segmentation / CNN features
Arbitrary Shape Text Detection via Segmentation with Probability Maps
Shi-Xue Zhang, Xiaobin Zhu, Lei Chen, Jie-Bo Hou, Xu-Cheng Yin
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 probability map supervision (PMS) (ablation) | Ours (no IM) · Total-Text | F-measure: 83.82 -> 86.09 | Improved |
| add probability map supervision (PMS) (ablation) | Ours (no IM) · MSRA-TD500 | F-measure: 81.07 -> 82.48 | Improved |
| add iterative module (IM) (ablation) | Ours (no PMS) · Total-Text | F-measure: 83.82 -> 84.68 | Improved |
| add iterative module (IM) (ablation) | Ours (no PMS) · MSRA-TD500 | F-measure: 81.07 -> 81.71 | Improved |
| add both PMS and IM (ablation) | Ours · Total-Text | F-measure: 83.82 -> 87.22 | Improved |
| add both PMS and IM (ablation) | Ours · MSRA-TD500 | F-measure: 81.07 -> 85.51 | Improved |
| add probability maps (speed effect) (ablation) | Ours · Total-Text | FPS: 5.6 -> 6.8 | Improved |
| increase iteration number n up to 4 (ablation) | Ours · Total-Text (no pretraining) | F-measure | Improved |
| use voting filtering instead of threshold filtering | Ours-1s (M+pse) · Total-Text | F-measure: 88.58 -> 88.72 | Improved |
| use voting filtering instead of threshold filtering | Ours-1s (M+pse) · CTW-1500 | F-measure: 85.41 -> 86.09 | Improved |
| use voting filtering instead of threshold filtering | Ours-1s (M+pse) · ICDAR2015 | F-measure: 87.25 -> 86.83 | Hurt |
| use watershed instead of PSE post-processing | Ours-1s (M) · Total-Text | F-measure: 88.58 -> 88.79 | Improved |
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.
- ContourNet — ContourNet: Taking a Further Step Toward Accurate Arbitrary-Shaped Scene Text Detection (CVPR 2020) · 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
- All You Need Is Boundary: Toward Arbitrary-Shaped Text Spotting (AAAI 2020 · similarity 0.90)
- Progressive Contour Regression for Arbitrary-Shape Scene Text Detection (CVPR 2021 · similarity 0.89)
- Efficient and Accurate Arbitrary-Shaped Text Detection With Pixel Aggregation Network (ICCV 2019 · similarity 0.89)
- Kernel Adaptive Convolution for Scene Text Detection via Distance Map Prediction (CVPR 2024 · similarity 0.88)
- LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network (AAAI 2024 · similarity 0.87)
- Towards Unified Multi-granularity Text Detection with Interactive Attention (ICML 2024 · 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.
