Research map / Detection and segmentation / CNN features
ContourNet: Taking a Further Step Toward Accurate Arbitrary-Shaped 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 |
|---|---|---|---|
| Use ContourNet with orthogonal texture modeling and Adaptive-RPN | Total-Text scene text detection · Total-Text | F-measure: 85.4 vs 83.5 | Helped |
| Use ContourNet with orthogonal texture modeling and Adaptive-RPN | CTW1500 scene text detection · CTW1500 | F-measure: 83.9 vs 82.2 | Helped |
| Use ContourNet instead of Wang et al. with SE blocks | ICDAR2015 scene text detection · ICDAR2015 | F-measure: 87.6 vs 86.9 | Hurt |
| Use 9 predefined points rather than 5 in Adaptive-RPN (ablation) | ContourNet · Total-Text | F-measure: 83.3 -> 83.9 | Helped |
| Replace conventional RPN with Adaptive-RPN (ablation) | ContourNet · Total-Text | F-measure: 83.1 -> 83.9 | Helped |
| Replace conventional RPN with Adaptive-RPN (ablation) | ContourNet · CTW1500 | F-measure: 84.5 -> 85.4 | Helped |
| Replace conventional RPN with Adaptive-RPN (ablation) | ContourNet · Total-Text small-size texts | F-measure gain: 1.4 | Helped |
| Replace conventional RPN with Adaptive-RPN (ablation) | ContourNet · Total-Text middle-size texts | F-measure gain: 0.3 | Helped |
| Replace conventional RPN with Adaptive-RPN (ablation) | ContourNet · Total-Text large-size texts | F-measure gain: 1.1 | Helped |
| Use LOTM with orthogonal texture modeling instead of a single-direction model (ablation) | ContourNet · Total-Text | F-measure: 80.6 -> 85.4 | Helped |
| Use orthogonal LOTM instead of joint 3 × 3 texture modeling (ablation) | ContourNet · Total-Text | F-measure: 84.0 -> 85.4 | Helped |
| Increase LOTM receptive-field size from 3 to 5 (ablation) | ContourNet · Total-Text | F-measure: 85.4 -> 84.7 | Hurt |
Closest papers by meaning
- Progressive Contour Regression for Arbitrary-Shape Scene Text Detection (CVPR 2021 · similarity 0.92)
- Arbitrary Shape Scene Text Detection With Adaptive Text Region Representation (CVPR 2019 · similarity 0.91)
- ABCNet: Real-Time Scene Text Spotting With Adaptive Bezier-Curve Network (CVPR 2020 · similarity 0.89)
- 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 0.88)
- CPN: Complementary Proposal Network for Unconstrained Text Detection (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.