Research map / Detection and segmentation / CNN features
Real-Time Scene Text Detection With Differentiable Binarization and Adaptive Scale Fusion
Minghui Liao, Zhisheng Zou, Zhaoyi Wan, Cong Yao, Xiang Bai
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 differentiable binarization (ablation) | segmentation text detector (ResNet-50) · MSRA-TD500 | F-measure: 78.7 -> 81.9 | Improved |
| add differentiable binarization (ablation) | segmentation text detector (ResNet-50) · CTW1500 | F-measure: 77.0 -> 81.6 | Improved |
| add differentiable binarization (ablation) | segmentation text detector (ResNet-18) · CTW1500 | F-measure: 74.5 -> 79.4 | Improved |
| add deformable convolution (ablation) | segmentation text detector (ResNet-50) · MSRA-TD500 | F-measure: 78.7 -> 83.7 | Improved |
| add adaptive scale fusion (ASF) (ablation) | segmentation text detector (ResNet-50) · MSRA-TD500 | F-measure: 78.7 -> 83.8 | Improved |
| add spatial attention inside ASF (ablation) | DBNet++ (ResNet-50) · MSRA-TD500 | F-measure: 86.7 -> 87.2 | Improved |
| add spatial attention inside ASF (ablation) | DBNet++ (ResNet-50) · CTW1500 | F-measure: 84.3 -> 85.3 | Improved |
| add supervision on the threshold map (ablation) | DBNet (ResNet-50) · MLT-2017 | F-measure: 72.1 -> 74.7 | Improved |
| add supervision on the threshold map (ablation) | DBNet (ResNet-18) · MLT-2017 | F-measure: 71.0 -> 71.7 | Improved |
| use ASF instead of Pyramid Pooling Module (ablation) | DBNet (ResNet-50) · MSRA-TD500 | F-measure: 85.1 -> 87.2 | Improved |
| use ASF instead of Criss-Cross Attention (ablation) | DBNet (ResNet-50) · MSRA-TD500 | F-measure: 86.5 -> 87.2 | Improved |
| add ASF to DBNet to get DBNet++ (ResNet-50) | DBNet (ResNet-50) · Total-Text | F-measure: 84.7 -> 86.0 | Improved |
Closest papers by meaning
- Real-Time Scene Text Detection with Differentiable Binarization (AAAI 2020 · similarity 0.95)
- MOST: A Multi-Oriented Scene Text Detector With Localization Refinement (CVPR 2021 · similarity 0.88)
- Kernel Adaptive Convolution for Scene Text Detection via Distance Map Prediction (CVPR 2024 · similarity 0.88)
- CBNet: A Plug-and-Play Network for Segmentation-Based Scene Text Detection (IJCV 2024 · similarity 0.87)
- Background-Insensitive Scene Text Recognition with Text Semantic Segmentation (ECCV 2022 · similarity 0.87)
- Towards Unified Multi-granularity Text Detection with Interactive Attention (ICML 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.
