Research map / Detection and segmentation / CNN features
Accurate Structured-Text Spotting for Arithmetical Exercise Correction
Yiqing Hu, Yan Zheng, Hao Liu, Dequang Jiang, Yinsong Liu, Bo Ren
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 |
|---|---|---|---|
| replace CenterNet loss with horizontal-focal loss (alpha=2) (ablation) | AEC detection · AEC-5k test (detection) | F-measure: 91.79 -> 93.04 | Improved |
| horizontal-focal loss alpha 2 -> 3 (ablation) | AEC detection · AEC-5k test (detection) | F-measure: 93.04 -> 89.35 | Hurt |
| horizontal-focal loss alpha 2 -> 5 (ablation) | AEC detection · AEC-5k test (detection) | F-measure: 93.04 -> 81.24 | Hurt |
| add multi-scale input {0.8,1,1.2} (ablation) | AEC detection · AEC-5k test (detection) | F-measure: 93.04 -> 95.76 | Improved |
| use MD-LSTM encoder instead of bi-LSTM (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 90.43 -> 91.11 | Improved |
| add context gate (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 90.71 -> 91.11 | Improved |
| use arithmetical attention instead of global attention (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 90.53 -> 91.11 | Improved |
| add resized candidates (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 91.11 -> 92.16 | Improved |
| add padded candidates (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 91.11 -> 91.83 | Improved |
| add rotated candidates (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 91.11 -> 91.84 | Improved |
| use candidates from all three actions (ablation) | ANMT · AEC-5k test (text spotting) | correction accuracy: 91.11 -> 93.72 | Improved |
| use all candidates (inference cost) (ablation) | ANMT · AEC-5k test (text spotting) | inference speed: 27% | 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.
- TextBoxes — TextBoxes: A Fast Text Detector with a Single Deep Neural Network (AAAI 2017) · better on 3 of 3 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
- MathMistake Checker: A Comprehensive Demonstration for Step-by-Step Math Problem Mistake Finding by Prompt-Guided LLMs (AAAI 2025 · similarity 0.83)
- Image Over Text: Transforming Formula Recognition Evaluation with Character Detection Matching (CVPR 2025 · similarity 0.80)
- DARR: A Dual-Branch Arithmetic Regression Reasoning Framework for Solving Machine Number Reasoning (AAAI 2025 · similarity 0.79)
- Uni-MuMER: Unified Multi-Task Fine-Tuning of Vision-Language Model for Handwritten Mathematical Expression Recognition (NeurIPS 2025 · similarity 0.79)
- Bridging the Gap Between End-to-End and Two-Step Text Spotting (CVPR 2024 · similarity 0.79)
- General Detection-based Text Line Recognition (NeurIPS 2024 · similarity 0.79)
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.
