Awesome
OCR Datasets
This repo collects OCR-related datasets. In general, the datasets are classified by 6 types, i.e., Natural Scene Text, Document Text, Handwritten Text, Historical Document Text, Video Text, and Synthetic Text.
<div align="center">
</div>
- Natural Scene Text: The images in this type of dataset are usually taken in natural scenes, so the difficulty of this task lies in the complex lighting transformations, shooting angles, blurring, varied fonts, etc.
- Document Text: only focues on document images, the difficulty is the variety of typesetting.
- Historical Document Text: is usally designed for assisting social science research. For example, digitized antiquarian documents help preserve historical materials and facilitate scholars to conduct related research.
- Video Text: aims at recognizing texts in videos, which introduces temporal information into the OCR task.
- Synthetic Text: synthetically generates images containing texts and the corresponding annotations by rendering texts of different fonts into natural photos. This type of dataset usually includes hundreds of thousands of samples since it does not require human beings to annotate the images. However, due to the limited technology, there is usually a large domain gap between the synthetic images and authentic samples; these datasets are often employed for pre-training only.
<table>
<thead>
<tr>
<th colspan="13">Natural Scene Text</th>
</tr>
</thead>
<tbody>
<tr>
<td colspan="2">Year/Venue</td>
<td>Name</td>
<td>Task</td>
<td>#Train(#wds)</td>
<td>#Val(#wds)</td>
<td>#Test(#wds)</td>
<td>Granu.</td>
<td>Anno. Form</td>
<td>Language</td>
<td>Scene</td>
<td>Paper</td>
<td>Size</td>
</tr>
<tr>
<td colspan="2">2003-05/ICDAR</td>
<td><a href="http://www.iapr-tc11.org/mediawiki/index.php?title=ICDAR_2003_Robust_Reading_Competitions" target="_blank" rel="noopener noreferrer">IC03/IC05</a></td>
<td>Det. & Rec.</td>
<td>258 (1110)</td>
<td>N/A</td>
<td>251 (1156)</td>
<td>Word</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>English</td>
<td>Natural</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=1227749" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>112MB</td>
</tr>
<tr>
<td colspan="2">2011-15/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=1" target="_blank" rel="noopener noreferrer">Born-DIgital-Image (IC2011-2015)</a></td>
<td>Det. & Rec. & Seg.</td>
<td>410 (3564)</td>
<td>N/A</td>
<td>141 (1439)</td>
<td>Word & Pixel</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>English</td>
<td>Natural/Web/Email</td>
<td><a href="http://www.cvc.uab.es/icdar2011competition/images/Report_RobustReading_Challenge1_final.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>40MB</td>
</tr>
<tr>
<td colspan="2">2013-15/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=2" target="_blank" rel="noopener noreferrer">Focused Scene Text (IC13)</a></td>
<td>Det. & Rec. & Seg.</td>
<td>229 (848)</td>
<td>N/A</td>
<td>233 (1095)</td>
<td>Word & Pixel</td>
<td>Rect [x1, y1, x2, y2, "transcript"] & SegMap</td>
<td>English</td>
<td>Natural</td>
<td><a href="http://dagdata.cvc.uab.es/icdar2013competition/files/icdar2013_competition_report.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>250MB</td>
</tr>
<tr>
<td colspan="2">2015/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=4" target="_blank" rel="noopener noreferrer">Incidental Scene Text (IC15)</a></td>
<td>Det. & Rec.</td>
<td>1,000 (4468)</td>
<td>N/A</td>
<td>500 (2077)</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Natural</td>
<td><a href="https://rrc.cvc.uab.es/files/short_rrc_2015.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>130MB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=8&com=introduction" target="_blank" rel="noopener noreferrer">Multi-Lingual Scene Text (MLT2017)</a></td>
<td>Det. & Rec.</td>
<td>7,200</td>
<td>1,800</td>
<td>private</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, Lan, 'trans']</td>
<td>multi-lingual</td>
<td>Natural</td>
<td>-</td>
<td>12GB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=15&com=introduction" target="_blank" rel="noopener noreferrer">Multi-Lingual Scene Text (MLT2019)</a></td>
<td>Det. & Rec.</td>
<td>10,000</td>
<td>N/A</td>
<td>10,000</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, Lan, 'trans']</td>
<td>multi-lingual</td>
<td>Natural</td>
<td><a href="https://arxiv.org/pdf/1907.00945.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~12GB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://bgshih.github.io/cocotext/" target="_blank" rel="noopener noreferrer">COCO-Text v2.0</a></td>
<td>Det. & Rec.</td>
<td>43,686</td>
<td>10,000</td>
<td>10,000</td>
<td>Word</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>En & NonEn</td>
<td>Natural</td>
<td><a href="https://vision.cornell.edu/se3/wp-content/uploads/2019/01/ICDAR2017b.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>13GB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=12&com=downloads" target="_blank" rel="noopener noreferrer">ReCTS</a></td>
<td>Det. & Rec.</td>
<td>20,000</td>
<td>N/A</td>
<td>5,000</td>
<td>Word/Line</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>Chinese</td>
<td>Signboard</td>
<td>-</td>
<td>~2.5GB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://github.com/cs-chan/Total-Text-Dataset" target="_blank" rel="noopener noreferrer">Total-Text</a></td>
<td>Det. & Rec.</td>
<td>1255</td>
<td>N/A</td>
<td>300</td>
<td>Word & Pixel</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>English</td>
<td>Natural</td>
<td><a href="https://arxiv.org/pdf/1710.10400.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>441MB</td>
</tr>
<tr>
<td colspan="2">2019/PR</td>
<td><a href="https://github.com/Yuliang-Liu/Curve-Text-Detector" target="_blank" rel="noopener noreferrer">SCUT-CTW1500</a></td>
<td>Det. & Rec.</td>
<td>1,000</td>
<td>N/A</td>
<td>500</td>
<td>Line</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>En & Ch</td>
<td>Natural</td>
<td><a href="https://arxiv.org/pdf/1712.02170.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>800MB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=14&com=introduction" target="_blank" rel="noopener noreferrer">Arbitrary-Shaped Text (ART)</a></td>
<td>Det. & Rec.</td>
<td>5,603 (50,029)</td>
<td>N/A</td>
<td>4,563 (52,631)</td>
<td>Word(En)/Line(CH)</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], Lan, 'trans']</td>
<td>En & Ch</td>
<td>Natural</td>
<td>-</td>
<td>4.4GB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="http://rctw.vlrlab.net/dataset/" target="_blank" rel="noopener noreferrer">RCTW-17 (CTW-12k)</a></td>
<td>Det. & Rec.</td>
<td>11514</td>
<td>N/A</td>
<td>1000</td>
<td>Line</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>Chinese</td>
<td>Mixture</td>
<td><a href="https://arxiv.org/pdf/1708.09585.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>11GB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR/ICCV</td>
<td><a href="https://rrc.cvc.uab.es/?ch=16" target="_blank" rel="noopener noreferrer">Large-scale Street View Text (LSVT)</a></td>
<td>Det. & Rec.</td>
<td>30,000</td>
<td>N/A</td>
<td>20,000</td>
<td>Line</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>En & Ch</td>
<td>Street View</td>
<td><a href="https://openaccess.thecvf.com/content_ICCV_2019/papers/Sun_Chinese_Street_View_Text_Large-Scale_Chinese_Text_Reading_With_Partially_ICCV_2019_paper.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>14GB</td>
</tr>
<tr>
<td colspan="2">2016/DAS</td>
<td><a href="https://github.com/lluisgomez/script_identification" target="_blank" rel="noopener noreferrer">MLe2e</a></td>
<td>Det. & Script Identifica.</td>
<td>450</td>
<td>N/A</td>
<td>261</td>
<td>Word</td>
<td>Rect [x1, y1, x2, y2, language] </td>
<td>multi-lingual</td>
<td>Natural</td>
<td><a href="https://arxiv.org/pdf/1602.07480.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>82MB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://cvit.iiit.ac.in/research/projects/cvit-projects/iiit-ilst" target="_blank" rel="noopener noreferrer">IIIT-ILST</a></td>
<td>Det. & Rec.</td>
<td>893</td>
<td></td>
<td></td>
<td>Word</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>Indic</td>
<td>Google Images</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8270315" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>609MB</td>
</tr>
<tr>
<td colspan="2">2017/CVPRW</td>
<td><a href="https://s3-us-west-2.amazonaws.com/uber-common-public/ubertext/index.html" target="_blank" rel="noopener noreferrer">UberText</a></td>
<td>Det. & Rec.</td>
<td>117,969 (571,534)</td>
<td></td>
<td></td>
<td>Word</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>English</td>
<td>Street View</td>
<td><a href="http://sunw.csail.mit.edu/abstract/uberText.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>197GB</td>
</tr>
<tr>
<td colspan="2">2009/VISAPP</td>
<td><a href="http://www.ee.surrey.ac.uk/CVSSP/demos/chars74k/" target="_blank" rel="noopener noreferrer">Chars74k</a></td>
<td>Det. & Rec.</td>
<td>1922</td>
<td></td>
<td></td>
<td>Character</td>
<td></td>
<td>En & Kanada</td>
<td>Natural Scene</td>
<td><a href="http://personal.ee.surrey.ac.uk/Personal/T.Decampos/papers/decampos_etal_visapp2009.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>739MB</td>
</tr>
<tr>
<td colspan="2">2010/ICPR</td>
<td><a href="http://www.iapr-tc11.org/mediawiki/index.php/KAIST_Scene_Text_Database" target="_blank" rel="noopener noreferrer">KAIST</a></td>
<td>Det. & Rec. & Seg.</td>
<td>3000</td>
<td></td>
<td></td>
<td>Char & Word & Pixel</td>
<td>Rect [x, y, w, h, "transcript"] & SegMap</td>
<td>En & Korean</td>
<td>Mixture</td>
<td><a href="http://milab.snu.ac.kr/pub/ICPR2010.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>364MB</td>
</tr>
<tr>
<td colspan="2">2010/ECCV</td>
<td><a href="http://vision.ucsd.edu/~kai/svt/" target="_blank" rel="noopener noreferrer">SVT</a></td>
<td>Det. & Rec.</td>
<td>100 (211)</td>
<td>N/A</td>
<td>250 (514)</td>
<td>Word</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>English</td>
<td>Street View</td>
<td><a href="http://vision.ucsd.edu/~kai/pubs/wang_eccv2010.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>118MB</td>
</tr>
<tr>
<td colspan="2">2013/ICCV</td>
<td><a href="https://pan.baidu.com/s/1rhYUn1mIo8OZQEGUZ9Nmrg" target="_blank" rel="noopener noreferrer">SVTP (download code:vnis)</a></td>
<td>Rec.</td>
<td>238 (639)</td>
<td></td>
<td></td>
<td>-</td>
<td></td>
<td>English</td>
<td>Street View</td>
<td><a href="https://ieeexplore.ieee.org/document/6751180/" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~1MB</td>
</tr>
<tr>
<td colspan="2">2011/NIPSw</td>
<td><a href="http://ufldl.stanford.edu/housenumbers/" target="_blank" rel="noopener noreferrer">SVHN</a></td>
<td>Det. & Rec.</td>
<td>73,257+531,131</td>
<td>N/A</td>
<td>26,032</td>
<td>Character</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>Digit</td>
<td>House Number</td>
<td><a href="https://storage.googleapis.com/pub-tools-public-publication-data/pdf/37648.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~3GB</td>
</tr>
<tr>
<td colspan="2">2011/ICDARw</td>
<td><a href="http://www.iapr-tc11.org/mediawiki/index.php?title=NEOCR:_Natural_Environment_OCR_Dataset" target="_blank" rel="noopener noreferrer">NEOCR</a></td>
<td>Det.</td>
<td>659 (5,238)</td>
<td></td>
<td></td>
<td>Line</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>multi-lingual</td>
<td>Natural Scene</td>
<td><a href="http://www.iapr-tc11.org/dataset/NEOCR/cbdar_paper.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>1.3GB</td>
</tr>
<tr>
<td colspan="2">2012/CVPR</td>
<td><a href="http://pages.ucsd.edu/%7Eztu/publication/MSRA-TD500.zip" target="_blank" rel="noopener noreferrer">MSRA-TD500</a></td>
<td>Det.</td>
<td>300</td>
<td>N/A</td>
<td>200</td>
<td>Line</td>
<td>RotRect [ind, difficult, x, y, w, h, theta]</td>
<td>multi-lingual</td>
<td>Street View</td>
<td><a href="https://pages.ucsd.edu/~ztu/publication/cvpr12_textdetection.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>96MB</td>
</tr>
<tr>
<td colspan="2">2012/BMVC</td>
<td><a href="http://pages.ucsd.edu/%7Eztu/publication/MSRA-TD500.zip" target="_blank" rel="noopener noreferrer">IIIT 5k-word</a></td>
<td>Rec.</td>
<td>380 (2000)</td>
<td>N/A</td>
<td>740 (3000)</td>
<td>Word</td>
<td></td>
<td>English</td>
<td>Natural</td>
<td><a href="http://cvit.iiit.ac.in/projects/SceneTextUnderstanding/IIIT5K.html" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>106MB</td>
</tr>
<tr>
<td colspan="2">2014/ESWA</td>
<td><a href="http://cs-chan.com/downloads_CUTE80_dataset.html" target="_blank" rel="noopener noreferrer">CUTE80</a></td>
<td>Rec.</td>
<td>80</td>
<td></td>
<td></td>
<td>Line</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]]]</td>
<td>English</td>
<td>Street View</td>
<td><a href="http://cs-chan.com/doc/ESWA_2014A.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>44MB</td>
</tr>
<tr>
<td colspan="2">2015/TPAMI</td>
<td><a href="http://prir.ustb.edu.cn/TexStar/MOMV-text-detection/" target="_blank" rel="noopener noreferrer">USTB-SV1K</a></td>
<td>Det. & Rec.</td>
<td>500</td>
<td>N/A</td>
<td>500</td>
<td>Word</td>
<td>RotRect [ind, difficult, x, y, w, h, theta, "trans"]</td>
<td>English</td>
<td>Street View</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7001081" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>36MB</td>
</tr>
<tr>
<td colspan="2">2019/JCST</td>
<td><a href="https://ctwdataset.github.io/" target="_blank" rel="noopener noreferrer">Chinese Text in the Wild (CTW)</a></td>
<td>Det. & Rec.</td>
<td>25,887(812,872chrs)</td>
<td>N/A</td>
<td>3,269(103,519chrs)</td>
<td>Char & Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>Chinese</td>
<td>Street View</td>
<td><a href="https://arxiv.org/pdf/1803.00085.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~40GB</td>
</tr>
<tr>
<td colspan="2">2019/TITS</td>
<td><a href="https://github.com/chongshengzhang/shopsign" target="_blank" rel="noopener noreferrer">ShopSign</a></td>
<td>Det. & Rec.</td>
<td>1258 sample images</td>
<td></td>
<td></td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>Chinese</td>
<td>Signboard</td>
<td><a href="https://ieeexplore.ieee.org/document/9186709" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>3GB</td>
</tr>
<tr>
<td colspan="2">2021/CVPR</td>
<td><a href="https://textvqa.org/textocr" target="_blank" rel="noopener noreferrer">TextOCR</a></td>
<td>Det. & Rec. & VQA</td>
<td>24902 (822,572)</td>
<td>N/A</td>
<td>3232 (80,497)</td>
<td>Word</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>English</td>
<td>Natural Scene</td>
<td><a href="https://arxiv.org/pdf/2105.05486.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~8GB</td>
</tr>
<tr>
<td colspan="2">2021/CVPR</td>
<td><a href="https://github.com/VinAIResearch/dict-guided#dataset" target="_blank" rel="noopener noreferrer">VinText</a></td>
<td>Det. & Rec.</td>
<td>1,200</td>
<td>N/A</td>
<td>300+500</td>
<td>Word</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>Vietnamese</td>
<td>Natural Scene</td>
<td><a href="https://www3.cs.stonybrook.edu/~minhhoai/papers/vintext_CVPR21.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>1GB</td>
</tr>
<tr>
<td colspan="2">2018/Competition</td>
<td><a href="https://tianchi.aliyun.com/competition/entrance/231685/introduction" target="_blank" rel="noopener noreferrer">ICPR MTWI2018</a></td>
<td>Det. & Rec.</td>
<td>10,000</td>
<td>N/A</td>
<td>10,000</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>En & Ch</td>
<td>WEB Images</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8546143" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>2GB</td>
</tr>
<tr>
<td colspan="2">2019/Competition</td>
<td><a href="https://aistudio.baidu.com/aistudio/competition/detail/20" target="_blank" rel="noopener noreferrer">百度中文场景文字识别比赛</a></td>
<td>Rec.</td>
<td>50,000</td>
<td>N/A</td>
<td>10,000</td>
<td>-</td>
<td>[h, w, 'trans']</td>
<td>En & Ch</td>
<td>Street View</td>
<td>-</td>
<td></td>
</tr>
<tr>
<td colspan="13">Document Text</td>
</tr>
<tr>
<td colspan="2">Year/Venue</td>
<td>Name</td>
<td>Task</td>
<td>#Train</td>
<td>#Val</td>
<td>#Test</td>
<td>Granu.</td>
<td>Anno. Form</td>
<td>Language</td>
<td>Scene</td>
<td>Paper</td>
<td>Size</td>
</tr>
<tr>
<td colspan="2">2011/ICDAR</td>
<td><a href="http://ciir.cs.umass.edu/downloads/ocr-evaluation/" target="_blank" rel="noopener noreferrer">RETAS</a></td>
<td colspan="4">No public download link </td>
<td>Char & Word</td>
<td>No public download link</td>
<td></td>
<td></td>
<td>-</td>
<td></td>
</tr>
<tr>
<td colspan="2">2013/IJDAR</td>
<td><a href="https://www.lrde.epita.fr/wiki/Olena/DatasetDBD" target="_blank" rel="noopener noreferrer">LRDE-DBD Document Binarization</a></td>
<td>Det. & Binarization</td>
<td>125</td>
<td></td>
<td></td>
<td>Line & Mask</td>
<td>Rect</td>
<td>French</td>
<td>Magzine</td>
<td><a href="https://www.lrde.epita.fr/wiki/Olena/DatasetDBD" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~700MB</td>
</tr>
<tr>
<td colspan="2">2015/ICDAR</td>
<td><a href="http://smartdoc.univ-lr.fr/smartdoc-2015-challenge-2-mobile-ocr-competition/smartdoc-2015-challenge-2-dataset/" target="_blank" rel="noopener noreferrer">SmartDOC</a></td>
<td></td>
<td>3630</td>
<td>N/A</td>
<td>8470</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td><a href="http://www.cvc.uab.es/~marcal/pdfs/ICDAR15e.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~30GB</td>
</tr>
<tr>
<td colspan="2">2016/ICFHR</td>
<td><a href="https://github.com/rahmad77/KPTI" target="_blank" rel="noopener noreferrer">KPTI</a></td>
<td>Rec.</td>
<td>11,910</td>
<td>2,552</td>
<td>2,553</td>
<td>-</td>
<td>['transcripts']</td>
<td>Pashto</td>
<td>Document</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=7814106" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~100MB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=9&com=introduction" target="_blank" rel="noopener noreferrer">DeText</a></td>
<td>Det. & Rec.</td>
<td>100</td>
<td>100</td>
<td>300</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Scientific<br></td>
<td><a href="https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0126200" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>10MB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=13" target="_blank" rel="noopener noreferrer">SROIE</a></td>
<td>Det. & Rec. & Info Ext.</td>
<td>600</td>
<td></td>
<td>400</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Receipt</td>
<td>-</td>
<td><1GB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://guillaumejaume.github.io/FUNSD/" target="_blank" rel="noopener noreferrer">FUNSD</a></td>
<td>Det. & Rec. & Info Ext.</td>
<td>149</td>
<td>N/A</td>
<td>50</td>
<td>Word</td>
<td>Rect [x1, y1, x2, y2, "transcript"]</td>
<td>English</td>
<td>Form</td>
<td><a href="https://arxiv.org/pdf/1905.13538.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>16MB</td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://github.com/herobd/NAF_dataset" target="_blank" rel="noopener noreferrer">NAF</a></td>
<td>Det. & Rec. & Info Ext.</td>
<td>682</td>
<td>59</td>
<td>63</td>
<td>Line</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Form</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=8977962" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2020</td>
<td><a href="https://github.com/ricardobnjunior/Brazilian-Identity-Document-Dataset" target="_blank" rel="noopener noreferrer">BID</a></td>
<td>Det. & Rec.</td>
<td>28880</td>
<td></td>
<td></td>
<td>Line</td>
<td>Poly</td>
<td>Latin</td>
<td>ID Document</td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">2020/ISCSIC</td>
<td><a href="https://github.com/machine-intelligence-laboratory/DDI-100" target="_blank" rel="noopener noreferrer">DDI-100</a></td>
<td>Det. & Rec.</td>
<td colspan="2">~ 100,000 (70% train, 30% val)</td>
<td></td>
<td>Char & Word & Mask</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Distorted Document</td>
<td><a href="https://arxiv.org/ftp/arxiv/papers/1912/1912.11658.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~300GB</td>
</tr>
<tr>
<td colspan="13">Handwritten Text</td>
</tr>
<tr>
<td colspan="2">Year/Venue</td>
<td>Name</td>
<td>Task</td>
<td>#Train</td>
<td>#Val</td>
<td>#Test</td>
<td>Granu.</td>
<td>Anno. Form</td>
<td>Language</td>
<td>Scene</td>
<td>Paper</td>
<td>Size</td>
</tr>
<tr>
<td colspan="2">2008-11/ICDAR</td>
<td><a href="http://www.a2ialab.com/doku.php?id=rimes_database:start" target="_blank" rel="noopener noreferrer">RIMES</a></td>
<td colspan="4">No public download link</td>
<td>Word & Line</td>
<td colspan="5">No public download link</td>
</tr>
<tr>
<td colspan="2">2010/DAS</td>
<td><a href="http://www.iapr-tc11.org/mediawiki/index.php/Harbin_Institute_of_Technology_Opening_Recognition_Corpus_for_Chinese_Characters_(HIT-OR3C)" target="_blank" rel="noopener noreferrer">HIT-OR3C</a></td>
<td>Rec.</td>
<td colspan="3">Char set 832,650 chars / Doc set 77,168 chars</td>
<td>-</td>
<td>special format</td>
<td>Chinese</td>
<td>Handwritten</td>
<td><a href="https://dl.acm.org/doi/pdf/10.1145/1815330.1815359" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>1GB</td>
</tr>
<tr>
<td colspan="2">2012/PR</td>
<td><a href="http://khatt.ideas2serve.net/index.php" target="_blank" rel="noopener noreferrer">KHATT</a></td>
<td>Rec.</td>
<td>8,368</td>
<td>1,793</td>
<td>1,822</td>
<td>-</td>
<td>['transcripts']</td>
<td>Arabic</td>
<td>Handwritten</td>
<td><a href="https://www.sciencedirect.com/science/article/pii/S0031320313003300" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">98-2014</td>
<td><a href="http://web.tuat.ac.jp/~nakagawa/database/" target="_blank" rel="noopener noreferrer">HANDS</a></td>
<td colspan="6">No public download link</td>
<td>Japanese</td>
<td>Handwritten</td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">-</td>
<td><a href="http://web.tuat.ac.jp/~nakagawa/database/Lao/abt.html" target="_blank" rel="noopener noreferrer">Lao-SABAIDEE</a></td>
<td>500 SAMPLES</td>
<td colspan="5">No public download link </td>
<td>Laos</td>
<td>Handwritten</td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">2014/ICFHR</td>
<td><a href="https://www.orand.cl/icfhr2014-hdsr/" target="_blank" rel="noopener noreferrer">ORAND-CAR/CVL</a></td>
<td>Rec.</td>
<td>5,000</td>
<td>N/A</td>
<td>5,000</td>
<td>Word</td>
<td>['image_name', 'trans']</td>
<td>Digits</td>
<td>Handwritten Digits</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6981115" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>194MB</td>
</tr>
<tr>
<td colspan="2">2018/ICFHR</td>
<td>VNOnDB</td>
<td>Rec.</td>
<td colspan="3">1,146 paragraphs 7,296 lines<br>380,000 chars</td>
<td>Word/Line/Parag.</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'trans']</td>
<td>Vietnamese</td>
<td>Handwritten</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8583810" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>200MB</td>
</tr>
<tr>
<td colspan="2">2013-16/IJDAR</td>
<td><a href="https://github.com/callee2006/HangulDB" target="_blank" rel="noopener noreferrer">PE92/SERI95/HanDB (HangulDB)</a></td>
<td>Rec.</td>
<td colspan="3">1200 samples (90% Train/10% Test)</td>
<td></td>
<td>.HGU1 format</td>
<td>Korean</td>
<td>Handwritten</td>
<td><a href="https://link.springer.com/article/10.1007/s10032-014-0229-4" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>800MB</td>
</tr>
<tr>
<td colspan="2">95-2016</td>
<td><a href="https://www.nist.gov/srd/nist-special-database-19" target="_blank" rel="noopener noreferrer">NIST</a></td>
<td>Rec.</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>English</td>
<td></td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">2011/ICDAR</td>
<td><a href="http://www.nlpr.ia.ac.cn/databases/handwriting/Home.html" target="_blank" rel="noopener noreferrer">CASIA-OLHWDB/HWDB</a></td>
<td>Rec.</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Chinese</td>
<td>Handwritten</td>
<td><a href="http://www.nlpr.ia.ac.cn/databases/download/ICDAR2011-CASIA%20databases.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="http://cvit.iiit.ac.in/research/projects/cvit-projects/iiit-indic-hw-words" target="_blank" rel="noopener noreferrer">IIT-INDIC-HW-WORDS</a></td>
<td>Rec.</td>
<td>872,000 instances</td>
<td></td>
<td></td>
<td>Word</td>
<td>['image_name', 'vocab_id'] & vocabularly</td>
<td>Indic</td>
<td>Handwritten</td>
<td><a href="http://cvit.iiit.ac.in/images/ConferencePapers/2021/iiit-indic-hw-words.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~20GB</td>
</tr>
<tr>
<td colspan="2">1999/ICDAR</td>
<td><a href="https://fki.tic.heia-fr.ch/databases/iam-handwriting-database" target="_blank" rel="noopener noreferrer">IAM Handwriting Database</a></td>
<td>Rec.</td>
<td>6,161</td>
<td>900+940</td>
<td>1,861</td>
<td colspan="6">Registration is Required</td>
</tr>
<tr>
<td colspan="2">2005/ICDAR</td>
<td><a href="https://fki.tic.heia-fr.ch/databases/iam-on-line-handwriting-database" target="_blank" rel="noopener noreferrer">IAM ONLINE Handwritting Data</a></td>
<td>Rec.</td>
<td>86,272 word instances</td>
<td colspan="8">Registration is Required</td>
</tr>
<tr>
<td colspan="2">2018/ICDAR</td>
<td><a href="https://fki.tic.heia-fr.ch/databases/iam-online-document-database" target="_blank" rel="noopener noreferrer">IAM-MonDo</a></td>
<td>Rec.</td>
<td colspan="7">Registration is Required </td>
<td><a href="https://dl.acm.org/doi/pdf/10.1145/1815330.1815343" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2011-14/ICDAR</td>
<td><a href="http://www.iapr-tc11.org/mediawiki/index.php?title=CROHME:_Competition_on_Recognition_of_Online_Handwritten_Mathematical_Expressions" target="_blank" rel="noopener noreferrer">CHROME</a></td>
<td>Rec.</td>
<td>> 10,000 expressions</td>
<td></td>
<td></td>
<td>symbol & expression</td>
<td>inkml format, latex</td>
<td>Symbol</td>
<td>Mathematical</td>
<td><a href="https://hal.archives-ouvertes.fr/file/index/docid/865627/filename/ICDAR_2013_CROHME.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>58MB</td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://ufal.mff.cuni.cz/muscima" target="_blank" rel="noopener noreferrer">MUSICMA++</a></td>
<td>Rec.</td>
<td>140</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Symbol</td>
<td>Music Notation</td>
<td><a href="https://arxiv.org/abs/1703.04824" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2018/Access</td>
<td><a href="https://github.com/HCIILAB/SCUT-EPT_Dataset_Release" target="_blank" rel="noopener noreferrer">SCUT-EPT</a></td>
<td>Rec.</td>
<td>40,000</td>
<td>N/A</td>
<td>10,000</td>
<td></td>
<td></td>
<td>Chinese</td>
<td>Educational Doc.</td>
<td><a href="https://ieeexplore.ieee.org/document/8565866" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>1.08GB</td>
</tr>
<tr>
<td colspan="2">2020/ICFHR</td>
<td><a href="https://www.cs.bgu.ac.il/~berat/" target="_blank" rel="noopener noreferrer">HHD</a></td>
<td>Rec.</td>
<td>3965</td>
<td></td>
<td>1134</td>
<td></td>
<td></td>
<td>Hebrew</td>
<td></td>
<td><a href="https://www.cs.bgu.ac.il/~berat/papers/icfhr2020_the_hhd_dataset.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ArXiv</td>
<td><a href="https://github.com/facebookresearch/IMGUR5K-Handwriting-Dataset" target="_blank" rel="noopener noreferrer">IMGUR5K</a></td>
<td>Det. & Rec.</td>
<td>(~108,000)</td>
<td>(~13,000)</td>
<td>(~14,000)</td>
<td>Word</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>English</td>
<td>Handwritten</td>
<td><a href="https://arxiv.org/pdf/2106.08385.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>-</td>
</tr>
<tr>
<td colspan="2">2021/ArXiv</td>
<td><a href="https://arxiv.org/pdf/2101.07542.pdf" target="_blank" rel="noopener noreferrer">VML-MOC</a></td>
<td>Seg. & Rec.</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Hebrew</td>
<td></td>
<td><a href="https://arxiv.org/pdf/2101.07542.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="https://www.kaggle.com/c/bengaliai-cv19/data" target="_blank" rel="noopener noreferrer">Bengali</a></td>
<td>Rec.</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Bengali</td>
<td></td>
<td><a href="https://arxiv.org/abs/2010.00170" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="https://goodnotes.com/gnhk/" target="_blank" rel="noopener noreferrer">GNHK</a></td>
<td>Det. & Rec.</td>
<td>687</td>
<td></td>
<td></td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td></td>
<td><a href="https://link.springer.com/chapter/10.1007/978-3-030-86337-1_27" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="13">Historical Document Text</td>
</tr>
<tr>
<td colspan="2">Year/Venue</td>
<td>Name</td>
<td>Task</td>
<td>#Train</td>
<td>#Val</td>
<td>#Test</td>
<td>Granu.</td>
<td>Anno. Form</td>
<td>Language</td>
<td>Scene</td>
<td>Paper</td>
<td>Size</td>
</tr>
<tr>
<td colspan="2">2010-11/DAS</td>
<td><a href="https://fki.tic.heia-fr.ch/databases/iam-historical-document-database" target="_blank" rel="noopener noreferrer">IAM-HistDB</a></td>
<td>Rec.</td>
<td>127</td>
<td></td>
<td></td>
<td>Word & Line</td>
<td>['image_id', 'transcript']</td>
<td>En & Ger & Latin</td>
<td></td>
<td></td>
<td>>200mb</td>
</tr>
<tr>
<td colspan="2">2016/ICFHR</td>
<td><a href="https://www.prhlt.upv.es/contests/icfhr2016-kws/data.html" target="_blank" rel="noopener noreferrer">H-KWS (1. Botany 2. AK)</a></td>
<td>Det. & Rec.</td>
<td>1849</td>
<td>3734</td>
<td>N/A</td>
<td>Word & Line</td>
<td>Rect [x, y, w, h, "transcript"]</td>
<td>English</td>
<td></td>
<td><a href="https://ieeexplore.ieee.org/document/7814133" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2016/ICFHR</td>
<td><a href="https://zenodo.org/record/1297399#.YUFmxHvhUXU" target="_blank" rel="noopener noreferrer">READ</a></td>
<td>Registration is Required</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>German</td>
<td></td>
<td><a href="https://ieeexplore.ieee.org/document/7814136" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~600mb</td>
</tr>
<tr>
<td colspan="2">2017/ICFHR</td>
<td><a href="http://amadi.univ-lr.fr/ICDAR2017_Competition/index.php/dataset" target="_blank" rel="noopener noreferrer">Palm Leaf Manuscript</a></td>
<td>Det. & Rec.</td>
<td colspan="3">~19,000 Balinese + ~20,000 Khmer</td>
<td>Char</td>
<td>No public download link</td>
<td>Khmer</td>
<td>Palm Leaf</td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">2017/HIP</td>
<td><a href="https://github.com/donavaly/SleukRith-Set" target="_blank" rel="noopener noreferrer">SleukRith-Set</a></td>
<td>Det. & Rec.</td>
<td>658</td>
<td></td>
<td></td>
<td>Char & Word</td>
<td>Polygon [[[x1,y1], [x2,y2], ..., [xn, yn]], 'transcript']</td>
<td>Khmer</td>
<td>Palm Leaf</td>
<td><a href="https://dl.acm.org/doi/10.1145/3151509.3151510" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>1GB</td>
</tr>
<tr>
<td colspan="2">2019/NCA</td>
<td><a href="https://ardisdataset.github.io/ARDIS/" target="_blank" rel="noopener noreferrer">ARDIS</a></td>
<td>Rec.</td>
<td>10,000</td>
<td></td>
<td></td>
<td>Char & Word</td>
<td>['transcript']</td>
<td>Digits</td>
<td>Church Records</td>
<td><a href="https://link.springer.com/article/10.1007/s00521-019-04163-3" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="https://www.cs.bgu.ac.il/~berat/" target="_blank" rel="noopener noreferrer">Pinkas</a></td>
<td>Det. & Rec.</td>
<td></td>
<td></td>
<td></td>
<td>Word & Line</td>
<td></td>
<td>Hebrew</td>
<td>historical manuscripts</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8978129" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>~50MB</td>
</tr>
<tr>
<td colspan="2">2020/ICFHR</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Cuneiform</td>
<td></td>
<td><a href="https://patrec.cs.tu-dortmund.de/pubs/papers/Rusakov2020-TQX" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2020/ICFHR</td>
<td><a href="https://github.com/HCIILAB/MTHv2_Datasets_Release" target="_blank" rel="noopener noreferrer">MTHv2</a></td>
<td>Det. & Rec.</td>
<td>2,399</td>
<td>N/A</td>
<td>800</td>
<td>Char & Line</td>
<td></td>
<td>Chinese</td>
<td>Acient Book</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9257624" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>4.6GB</td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="https://morphoboid.labri.fr/ihr-nom.html" target="_blank" rel="noopener noreferrer">IHR-NomDB</a></td>
<td>Det. & Rec.</td>
<td>267</td>
<td></td>
<td></td>
<td>Line</td>
<td>Rect [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>ChuNom</td>
<td>Acient Book</td>
<td><a href="https://link.springer.com/chapter/10.1007/978-3-030-86334-0_6#Sec3" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="https://www.cs.bgu.ac.il/~berat/" target="_blank" rel="noopener noreferrer">VML-HP</a></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Hebrew</td>
<td></td>
<td><a href="https://link.springer.com/content/pdf/10.1007%2F978-3-030-86337-1_14.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2"></td>
<td><a href="https://www.cs.bgu.ac.il/~berat/" target="_blank" rel="noopener noreferrer">VML-AHTE</a></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td><a href="https://arxiv.org/pdf/2101.08299.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2019/ICDAR</td>
<td><a href="http://ihdia.iiit.ac.in/indiscapes/" target="_blank" rel="noopener noreferrer">IndiScapes</a></td>
<td>Seg</td>
<td>No public download link</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Indic</td>
<td></td>
<td><a href="https://arxiv.org/pdf/1912.07025.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="13">Video Text</td>
</tr>
<tr>
<td colspan="2">Year/Venue</td>
<td>Name</td>
<td>Task</td>
<td>#TrainVids (#frames)</td>
<td>#ValVids (#f)</td>
<td>#TestVids(#f)</td>
<td>Granu.</td>
<td>Anno. Form</td>
<td>Language</td>
<td>Scene</td>
<td>Paper</td>
<td>Size</td>
</tr>
<tr>
<td colspan="2">2013/15/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=3" target="_blank" rel="noopener noreferrer">Text in Videos (IC13)</a></td>
<td>Det. & Rec.</td>
<td>25 (13450)</td>
<td></td>
<td>24 (14374)</td>
<td>Word</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Natural</td>
<td><a href="http://dagdata.cvc.uab.es/icdar2013competition/files/icdar2013_competition_report.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2015/ICDAR</td>
<td><a href="http://www.ict.griffith.edu.au/cvsi2015/Dataset.php" target="_blank" rel="noopener noreferrer">CVSI2015</a></td>
<td colspan="6">No public link for download</td>
<td>multi-lingual</td>
<td></td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7333950" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2017/ICDAR</td>
<td><a href="https://rrc.cvc.uab.es/?ch=7" target="_blank" rel="noopener noreferrer">DOST</a></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Word</td>
<td>QUAD</td>
<td>Japanese</td>
<td></td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">2018/ICFHR</td>
<td><a href="https://cvit.iiit.ac.in/research/projects/cvit-projects/lecturevideodb" target="_blank" rel="noopener noreferrer">LectureVideoDB</a></td>
<td>Det. & Rec.</td>
<td>-52,225</td>
<td>-27,900</td>
<td>-36,460</td>
<td>Word</td>
<td></td>
<td>English</td>
<td>Slides/Paper</td>
<td><a href="https://ieeexplore.ieee.org/document/8583767" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>2.3GB</td>
</tr>
<tr>
<td colspan="2">2020/ICRA</td>
<td><a href="http://cvit.iiit.ac.in/research/projects/cvit-projects/roadtext-1k" target="_blank" rel="noopener noreferrer">RoadText-1K</a></td>
<td>Det. & Rec.</td>
<td>500 (150,000)</td>
<td>200 (60,000)</td>
<td>300 (90,000)</td>
<td>Line</td>
<td>Rect [x1, y1, x2, y2, "transcript"] & SegMap</td>
<td>En & NonEn</td>
<td>Road/Traffic</td>
<td><a href="https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=9196577" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2020/ICMV</td>
<td><a href="https://github.com/fcakyon/midv500" target="_blank" rel="noopener noreferrer">MIDV-500 & MIDV-2019</a></td>
<td>Det. & Rec. & Others</td>
<td>500 video clips</td>
<td></td>
<td></td>
<td></td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>multi-lingual</td>
<td>Document</td>
<td><a href="https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11433/2558438/MIDV-2019--challenges-of-the-modern-mobile-based-document/10.1117/12.2558438.full?SSO=1" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>32GB</td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="ftp://smartengines.com/midv-lait/" target="_blank" rel="noopener noreferrer">MIDV-LAIT</a></td>
<td>Det. & Rec. & Others</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>multi-lingual</td>
<td>Document</td>
<td><a href="https://link.springer.com/chapter/10.1007/978-3-030-86331-9_17#Sec3" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2020/ICPR</td>
<td><a href="https://diuf.unifr.ch/main/diva/AcTiVComp/evaluation.html" target="_blank" rel="noopener noreferrer">AcTiVComp</a></td>
<td>Det. & Rec.</td>
<td>2557 frames</td>
<td></td>
<td></td>
<td>Line</td>
<td>Rect [x1, y1, x2, y2, "transcript"]</td>
<td>Arabic</td>
<td></td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="13">Synthetic Text</td>
</tr>
<tr>
<td colspan="2">Year/Venue</td>
<td>Name</td>
<td>Task</td>
<td>#Train</td>
<td>#Val</td>
<td>#Test</td>
<td>Granu.</td>
<td>Anno. Form</td>
<td>Language</td>
<td>Scene</td>
<td>Paper</td>
<td>Size</td>
</tr>
<tr>
<td colspan="2">2016/CVPR</td>
<td><a href="https://www.robots.ox.ac.uk/~vgg/data/scenetext/" target="_blank" rel="noopener noreferrer">Synth800k</a></td>
<td>Det. & Rec.</td>
<td>858,750 (7,266,866)</td>
<td></td>
<td></td>
<td>Char & Word & Line</td>
<td>Quad [x1, y1, x2, y2, x3, y3, x4, y4, 'trans']</td>
<td>English</td>
<td>Synthetic</td>
<td><a href="https://arxiv.org/pdf/1604.06646.pdf" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td>41GB</td>
</tr>
<tr>
<td colspan="2">2020</td>
<td><a href="https://jyouhou.github.io/UnrealText/" target="_blank" rel="noopener noreferrer">UnrealText</a></td>
<td></td>
<td colspan="3">728,000 En + 674,000 others</td>
<td></td>
<td></td>
<td>multi-lingual</td>
<td></td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">-</td>
<td><a href="https://github.com/YCG09/chinese_ocr" target="_blank" rel="noopener noreferrer">Chinese_ocr</a></td>
<td>Det. & Rec.</td>
<td>~ 364 million</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Chinese</td>
<td>Document</td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">-</td>
<td><a href="https://tukl.seecs.nust.edu.pk/downloads.html" target="_blank" rel="noopener noreferrer">UPTI</a></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td>Urdu</td>
<td></td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">-</td>
<td><a href="https://diuf.unifr.ch/main/diva/APTI/" target="_blank" rel="noopener noreferrer">APTI</a></td>
<td></td>
<td colspan="3">45313600 (> 250 million chars)</td>
<td>Word</td>
<td></td>
<td>arabic</td>
<td></td>
<td></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="https://github.com/clovaai/synthtiger" target="_blank" rel="noopener noreferrer">SynthTiger</a></td>
<td>Rec.</td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td><a href="https://link.springer.com/chapter/10.1007/978-3-030-86337-1_8#Sec6" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
<tr>
<td colspan="2">2021/ICDAR</td>
<td><a href="https://github.com/biswassanket/synth_doc_generation" target="_blank" rel="noopener noreferrer">DocSynth</a></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td></td>
<td><a href="https://link.springer.com/chapter/10.1007/978-3-030-86334-0_36" target="_blank" rel="noopener noreferrer">PDF</a></td>
<td></td>
</tr>
</tbody>
</table>