Research map / Detection and segmentation / CNN features
PAN++: Towards Efficient and Accurate End-to-End Spotting of Arbitrarily-Shaped Text
Wenhai Wang, Enze Xie, Xiang Li, Xuebo Liu, Ding Liang, Zhibo Yang, Tong Lü, Chunhua Shen
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 |
|---|---|---|---|
| stack 1 FPEM instead of none (ablation) | PAN++ · Total-Text spotting | end-to-end F-measure: 64.5 -> 66.0 | Improved |
| stack 2 FPEMs instead of none (ablation) | PAN++ · Total-Text spotting | end-to-end F-measure: 64.5 -> 66.4 | Improved |
| stack 4 FPEMs instead of 2 (ablation) | PAN++ · Total-Text spotting | end-to-end F-measure: 66.4 -> 66.6 | No clear effect |
| stack 2 FPEMs instead of none (ablation) | PAN++ · IC15 spotting | end-to-end F-measure: 63.7 -> 64.7 | Improved |
| stack 4 FPEMs instead of 2 (ablation) | PAN++ · Total-Text spotting | FPEM time (ms): 3.8 -> 6.5 | Hurt |
| add improved Pixel Aggregation (background term) to PAN (ablation) | PAN · Total-Text detection | F-measure: 83.5 -> 84.0 | Improved |
| add improved FPEM to PAN + PA (impr) (ablation) | PAN · Total-Text detection | F-measure: 84.0 -> 84.5 | Improved |
| add improved Pixel Aggregation to PAN (ablation) | PAN · IC15 detection | F-measure: 80.3 -> 80.7 | Improved |
| add improved FPEM to PAN + PA (impr) (ablation) | PAN · IC15 detection | F-measure: 80.7 -> 81.2 | Improved |
| fine-tune on SynthText before Total-Text | PAN++ detection · Total-Text detection | F-measure: 84.5 -> 85.3 | Improved |
| joint training on multiple datasets instead of fine-tuning | PAN++ (S:736) · Total-Text spotting | end-to-end F-measure: 67.1 -> 68.6 | Improved |
| raise shrinking rate to 1 (kernel equals text region) (ablation) | PAN++ detection · Total-Text and IC15 | F-measure | Hurt |
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.
- ABCNet — ABCNet: Real-Time Scene Text Spotting With Adaptive Bezier-Curve Network (CVPR 2020) · better on 1 of 1 shared result
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
- LRANet++: Low-Rank Approximation Network for Accurate and Efficient Text Spotting (TPAMI 2026 · similarity 0.90)
- Towards Unconstrained End-to-End Text Spotting (ICCV 2019 · similarity 0.90)
- Efficient and Accurate Arbitrary-Shaped Text Detection With Pixel Aggregation Network (ICCV 2019 · similarity 0.90)
- All You Need Is Boundary: Toward Arbitrary-Shaped Text Spotting (AAAI 2020 · similarity 0.90)
- Towards End-to-End Text Spotting in Natural Scenes (TPAMI 2021 · similarity 0.89)
- Kernel Adaptive Convolution for Scene Text Detection via Distance Map Prediction (CVPR 2024 · similarity 0.88)
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.
