Research map / Detection and segmentation / CNN features
Progressive Contour Regression for Arbitrary-Shape Scene Text Detection
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 |
|---|---|---|---|
| Add oriented text proposal generation before contour regression (ablation) | PCR · CTW1500 | F-measure: 77.2 -> 79.8 | Helped |
| Add contour information aggregation (ablation) | PCR with OTPG · CTW1500 | F-measure: 79.8 -> 82.4 | Helped |
| Add reliable contour localization (ablation) | PCR with OTPG · CTW1500 | F-measure: 79.8 -> 80.5 | Helped |
| Add both contour information aggregation and reliable contour localization (ablation) | PCR with OTPG · CTW1500 | F-measure: 79.8 -> 83.7 | Helped |
| Add one iterative contour localization module (ablation) | PCR · CTW1500 | F-measure: 83.0 -> 83.7 | Helped |
| Increase iterative contour localization modules from 1 to 2 (ablation) | PCR · CTW1500 | F-measure: 83.7 -> 84.0 | Helped |
| Increase iterative contour localization modules from 2 to 3 (ablation) | PCR · CTW1500 | F-measure: 84.0 -> 83.8 | Hurt |
| Use 16 local sink nodes per CIA unit rather than 0 (ablation) | PCR · CTW1500 | F-measure: 83.2 -> 84.0 | Helped |
| Vary local sink nodes per CIA unit from 16 to 32 (ablation) | PCR · CTW1500 | F-measure: 84.0 -> 83.6 | Hurt |
| Replace CIA units with one circular convolution layer (ablation) | PCR · CTW1500 | F-measure: 84.0 -> 82.6 | Hurt |
| Replace circular convolution layers with standard convolution layers (ablation) | PCR · CTW1500 | F-measure: 84.0 -> 83.6 | Hurt |
| Use the proposed DLA-34 PCR configuration with external pretraining | PCR · CTW1500 | F-measure: 84.0 -> 84.7 | Helped |
Closest papers by meaning
- ContourNet: Taking a Further Step Toward Accurate Arbitrary-Shaped Scene Text Detection (CVPR 2020 · similarity 0.92)
- LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network (AAAI 2024 · similarity 0.89)
- Kernel Adaptive Convolution for Scene Text Detection via Distance Map Prediction (CVPR 2024 · similarity 0.86)
- LayoutFormer: Hierarchical Text Detection Towards Scene Text Understanding (CVPR 2024 · similarity 0.85)
- Towards Unified Multi-granularity Text Detection with Interactive Attention (ICML 2024 · similarity 0.85)
- CPN: Complementary Proposal Network for Unconstrained Text Detection (AAAI 2024 · similarity 0.84)
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.