Research map / Detection and segmentation / CNN features
Geometry Normalization Networks for Accurate Scene Text Detection
Youjiang Xu, Jiaqi Duan, Zhanghui Kuang, Xiaoyu Yue, Hongbin Sun, Yue Guan, Wayne Zhang
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 the second scale branch S1/2 to the single-branch SNU (ablation) | EAST · ICDAR 2015 | F-score: 80.63 -> 84.81 | Improved |
| add a third scale branch S1/4 (ablation) | EAST + SNU · ICDAR 2015 | F-score: 84.81 -> 83.16 | Hurt |
| add a third scale branch S1/4 (ablation) | PSENet + SNU · ICDAR 2015 | F-score: 85.95 -> 85.50 | Hurt |
| use four orientation branches instead of one (ablation) | EAST · Rotated ICDAR 2015 | F-score: 69.23 -> 76.42 | Improved |
| use four orientation branches instead of one (ablation) | PSENet · Rotated ICDAR 2015 | F-score: 77.41 -> 78.98 | Improved |
| add Init (initialize from baseline models) (ablation) | EAST GNNets · Rotated ICDAR 2015 | F-score: 67.11 -> 75.15 | Improved |
| add geometry-aware training sampling strategy (ablation) | EAST GNNets · Rotated ICDAR 2015 | F-score: 75.15 -> 80.82 | Improved |
| add rotation augmentation (ablation) | EAST · ICDAR 2015 | F-score: 80.63 -> 72.43 | Hurt |
| add SNU to ONU-only GNNets (ablation) | EAST · Rotated ICDAR 2015 | F-score: 76.42 -> 80.82 | Improved |
| add ONU to SNU-only GNNets (ablation) | EAST · Rotated ICDAR 2015 | F-score: 73.47 -> 80.82 | Improved |
| add GNM without Init and TSS (ablation) | PSENet+RA · Rotated ICDAR 2015 | F-score: 77.41 -> 75.29 | Hurt |
| replace PSENet with GNNets (SNU only) | PSENet (reimplemented) · ICDAR 2017 MLT | F-score: 72.74 -> 74.54 | 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.
- FOTS — FOTS: Fast Oriented Text Spotting With a Unified Network (CVPR 2018) · better on 2 of 2 shared results
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
- Geometry-Aware Scene Text Detection With Instance Transformation Network (CVPR 2018 · similarity 0.88)
- TextSnake: A Flexible Representation for Detecting Text of Arbitrary Shapes (ECCV 2018 · similarity 0.88)
- Text Grouping Adapter: Adapting Pre-trained Text Detector for Layout Analysis (CVPR 2024 · similarity 0.85)
- LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network (AAAI 2024 · similarity 0.85)
- LayoutFormer: Hierarchical Text Detection Towards Scene Text Understanding (CVPR 2024 · similarity 0.85)
- 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.
