Research map / Detection and segmentation / CNN features
Shape Robust Text Detection With Progressive Scale Expansion Network
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 multiple kernel scales (n=6, m=0.5) instead of one full-scale segmentation mask (ablation) | PSENet-1s · ICDAR 2015 test | F-measure: 68.56 -> 87.08 | Helped |
| Increase the number of kernel scales from 2 toward 6 while keeping m=0.5 (ablation) | PSENet-1s · ICDAR 2015 test | F-measure | Helped |
| Increase the number of kernel scales beyond 6 while keeping m=0.5 (ablation) | PSENet-1s · ICDAR 2015 test | F-measure | No clear effect |
| Choose a minimal scale m that is too large or too small (ablation) | PSENet-1s with n=6 · ICDAR 2015 test | F-measure | Hurt |
| Use 2s instead of 1s output-map scale (ablation) | PSENet · ICDAR 2015 test | F-measure: 87.08 -> 87.21 | Helped |
| Use 4s instead of 1s output-map scale (ablation) | PSENet · ICDAR 2015 test | F-measure: 87.08 -> 85.88 | Hurt |
| Use 2s instead of 1s output-map scale (ablation) | PSENet · ICDAR 2017 MLT test | F-measure: 72.45 -> 72.40 | No clear effect |
| Use 4s instead of 1s output-map scale (ablation) | PSENet · ICDAR 2017 MLT test | F-measure: 72.45 -> 71.52 | Hurt |
| Use 2s instead of 1s output-map scale (ablation) | PSENet · SCUT-CTW1500 test | F-measure: 81.17 -> 80.60 | Hurt |
| Use 4s instead of 1s output-map scale (ablation) | PSENet · SCUT-CTW1500 test | F-measure: 81.17 -> 79.29 | Hurt |
Closest papers by meaning
- Fourier Contour Embedding for Arbitrary-Shaped Text Detection (CVPR 2021 · similarity 0.88)
- Progressive Contour Regression for Arbitrary-Shape Scene Text Detection (CVPR 2021 · similarity 0.88)
- ContourNet: Taking a Further Step Toward Accurate Arbitrary-Shaped Scene Text Detection (CVPR 2020 · similarity 0.88)
- LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network (AAAI 2024 · similarity 0.88)
- Kernel Adaptive Convolution for Scene Text Detection via Distance Map Prediction (CVPR 2024 · similarity 0.87)
- CPN: Complementary Proposal Network for Unconstrained Text Detection (AAAI 2024 · similarity 0.85)
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.