icdar2013
Dataset from Papers With Code
Legacy benchmark from 2013. For current OCR evaluation, use OCRBench v2, ICDAR 2015, or newer benchmarks.
DTrOCR 105M
Unknown
99.4
accuracy
accuracy Progress Over Time
Showing 14 breakthroughs from Jun 2014 to Aug 2023
Key Milestones
From paper: Synthetic Data and Artificial Neural Networks for Natural Scene Text Recognition
From paper: An End-to-End Trainable Neural Network for Image-based Sequence Recognition and Its Application to Scene Text Recognition
From paper: Star-net: A spatial attention residue network for scene text recognition.
From paper: ASTER: An Attentional Scene Text Recognizer with Flexible Rectification
From paper: What Is Wrong With Scene Text Recognition Model Comparisons? Dataset and Model Analysis
From paper: On Recognizing Texts of Arbitrary Shapes with 2D Self-Attention
From paper: Towards Accurate Scene Text Recognition with Semantic Reasoning Networks
From paper: Why You Should Try the Real Data for the Scene Text Recognition
From paper: Look Back Again: Dual Parallel Attention Network for Accurate and Robust Scene Text Recognition
From paper: Multi-modal Text Recognition Networks: Interactive Enhancements between Visual and Semantic Features
From paper: CLIP4STR: A Simple Baseline for Scene Text Recognition with Pre-trained Vision-Language Model
From paper: DTrOCR: Decoder-only Transformer for Optical Character Recognition
Top Models Performance Comparison
Top 10 models ranked by accuracy
accuracyPrimary
| # | Model | Score | Paper / Code | Date |
|---|---|---|---|---|
| 1 | DTrOCR 105M | 99.4 | Aug 2023 | |
| 2 | CLIP4STR-L (DataComp-1B) | 99 | May 2023 | |
| 3 | MGP-STR | 98.5 | Sep 2022 | |
| 4 | CLIP4STR-L | 98.5 | May 2023 | |
| 5 | CLIP4STR-B* | 98.3 | May 2023 | |
| 6 | CCD-ViT-Base(ARD_2.8M) | 98.3 | Nov 2022 | |
| 7 | CCD-ViT-Small(ARD_2.8M) | 98.3 | Nov 2022 | |
| 8 | MATRN | 97.9 | Nov 2021 | |
| 9 | S-GTR | 97.8 | Dec 2021 | |
| 10 | SIGA_T | 97.8 | Mar 2022 | |
| 11 | DPAN | 97.7 | Look Back Again: Dual Parallel Attention Network for Accurate and Robust Scene Text RecognitionCode | Aug 2021 |
| 12 | CDistNet (Ours) | 97.67 | Nov 2021 | |
| 13 | CCD-ViT-Tiny(ARD_2.8M) | 97.5 | Nov 2022 | |
| 14 | SVTR-L (Large) | 97.2 | Apr 2022 | |
| 15 | SVTR-B (Base) | 97.1 | Apr 2022 | |
| 16 | DiffusionSTR | 97.1 | Jun 2023 | |
| 17 | Yet Another Text Recognizer | 96.8 | Jul 2021 | |
| 18 | SVTR-T (Tiny) | 96.3 | Apr 2022 | |
| 19 | SVTR-S (Small) | 95.7 | Apr 2022 | |
| 20 | SRN | 95.5 | Mar 2020 | |
| 21 | RCEED | 94.7 | Jun 2021 | |
| 22 | SATRN | 94.1 | Oct 2019 | |
| 23 | DAN | 93.9 | Dec 2019 | |
| 24 | CSTR | 93.2 | Feb 2021 | |
| 25 | TextScanner | 92.9 | Dec 2019 | |
| 26 | SAFL | 92.8 | Jan 2022 | |
| 27 | SEED | 92.8 | May 2020 | |
| 28 | ViTSTR | 92.4 | May 2021 | |
| 29 | Baek et al. | 92.3 | Apr 2019 | |
| 30 | ASTER | 91.8 | ASTER: An Attentional Scene Text Recognizer with Flexible RectificationCode | Jun 2018 |
| 31 | CA-FCN | 91.5 | Sep 2018 | |
| 32 | SAR | 91 | Nov 2018 | |
| 33 | STAR-Net | 89.1 | Star-net: A spatial attention residue network for scene text recognition.Code | Sep 2016 |
| 34 | RARE | 88.6 | Mar 2016 | |
| 35 | CRNN | 86.7 | Jul 2015 | |
| 36 | CHAR | 79.5 | Jun 2014 |
avg-f1
| # | Model | Score | Paper / Code | Date |
|---|---|---|---|---|
| 1 | CDeCNet | 1 | Aug 2020 | |
| 2 | cascadetabnet | 1 | Apr 2020 | |
| 3 | TableNet | 0.966 | Jan 2020 |