remiai3/synthetic-captchas-library

🌍 Synthetic Multilingual CAPTCHA Library

Repository: remiai3/synthetic-captchas-library A multilingual dataset of synthetic 4-character CAPTCHA images designed for OCR, multilingual vision models, and script recognition research. This dataset spans 44 world writing systems and is especially useful for low-resource script OCR training.

📌 Dataset Summary

Each script includes 100,000 unique CAPTCHA images. The dataset is provided in two parallel formats:

• —CSV version (standard ML workflows)
• —Parquet version (faster loading for large-scale pipelines)

⚠️ The images inside both folders are identical — only the label file format differs. So per language:

• —100,000 unique images
• —100,000 duplicate copies (same images, different label format)

🗂 Dataset Structure

tiny-captcha-library/ │ ├── 4char/ │ ├── <LanguageName>/ │ │ ├── images.zip │ │ └── labels.csv │ └── ... │ ├── 4charparquet/ │ ├── <Language_Name>/ │ │ ├── images.zip │ │ └── labels.parquet │ └── ... │ ├── Amharic.png ├── Armenian.png └── ...

Folder Details

Each language folder includes:

• —images.zip → CAPTCHA images
• —Label file mapping image filename → correct 4-character text

🌐 Supported Scripts

Amharic, Armenian, Arabic, Bengali, Cherokee, Chinese, Georgian, Greek, Gujarati, Hebrew, Hindi, Hanunoo, Japanese, Kaithi, Kannada, Khmer, Lao, Latin, Lisu, Malayalam, Miao, Modi, Myanmar, Odia, Osage, Russian, Sharada, Siddham, Sinhala, Soyombo, Tai Tham, Tai Viet, Takri, Thaana, Tirhuta, Tamil, Telugu, Thai, Ukrainian, Tifinagh, Korean, Tibetan, Ethiopian, Adlam

🖼 Sample CAPTCHA Styles

<table align="center"> <tr> <td align="center"><img src="./Amharic.png" width="110"/><br>Amharic</td> <td align="center"><img src="./Arabic.png" width="110"/><br>Arabic</td> <td align="center"><img src="./Armenian.png" width="110"/><br>Armenian</td> <td align="center"><img src="./Bengali.png" width="110"/><br>Bengali</td> <td align="center"><img src="./Cherokee.png" width="110"/><br>Cherokee</td> <td align="center"><img src="./Chinese.png" width="110"/><br>Chinese</td> </tr>

<tr> <td align="center"><img src="./Georgian.png" width="110"/><br>Georgian</td> <td align="center"><img src="./Greek.png" width="110"/><br>Greek</td> <td align="center"><img src="./Gujarati.png" width="110"/><br>Gujarati</td> <td align="center"><img src="./Hanunoo.png" width="110"/><br>Hanunoo</td> <td align="center"><img src="./Hebrew.png" width="110"/><br>Hebrew</td> <td align="center"><img src="./Hindi.png" width="110"/><br>Hindi</td> </tr>

<tr> <td align="center"><img src="./Japanesetiny.png" width="110"/><br>Japanese</td> <td align="center"><img src="./Kaithi.png" width="110"/><br>Kaithi</td> <td align="center"><img src="./Kannada.png" width="110"/><br>Kannada</td> <td align="center"><img src="./Khmer.png" width="110"/><br>Khmer</td> <td align="center"><img src="./Lao.png" width="110"/><br>Lao</td> <td align="center"><img src="./Latin.png" width="110"/><br>Latin</td> </tr>

<tr> <td align="center"><img src="./Lisu.png" width="110"/><br>Lisu</td> <td align="center"><img src="./Malayalam.png" width="110"/><br>Malayalam</td> <td align="center"><img src="./Miao.png" width="110"/><br>Miao</td> <td align="center"><img src="./Modi.png" width="110"/><br>Modi</td> <td align="center"><img src="./Myanmar.png" width="110"/><br>Myanmar</td> <td align="center"><img src="./Odia.png" width="110"/><br>Odia</td> </tr>

<tr> <td align="center"><img src="./Osage.png" width="110"/><br>Osage</td> <td align="center"><img src="./Russian.png" width="110"/><br>Russian</td> <td align="center"><img src="./Sharada.png" width="110"/><br>Sharada</td> <td align="center"><img src="./Siddham.png" width="110"/><br>Siddham</td> <td align="center"><img src="./Sinhala.png" width="110"/><br>Sinhala</td> <td align="center"><img src="./Soyombo.png" width="110"/><br>Soyombo</td> </tr>

<tr> <td align="center"><img src="./TaiTham.png" width="110"/><br>Tai Tham</td> <td align="center"><img src="./TaiViet.png" width="110"/><br>Tai Viet</td> <td align="center"><img src="./Takri.png" width="110"/><br>Takri</td> <td align="center"><img src="./Tamil.png" width="110"/><br>Tamil</td> <td align="center"><img src="./Telugu.png" width="110"/><br>Telugu</td> <td align="center"><img src="./Thaana.png" width="110"/><br>Thaana</td> </tr>

<tr> <td align="center"><img src="./Thai.png" width="110"/><br>Thai</td> <td align="center"><img src="./Tirhuta.png" width="110"/><br>Tirhuta</td> <td align="center"><img src="./Ukrainian.png" width="110"/><br>Ukrainian</td> <td align="center"><img src="./Tifinagh.png" width="110"/><br>Tifinagh</td> <td align="center"><img src="./Koreantiny.png" width="110"/><br>Korean</td> <td align="center"><img src="./Tibetan.png" width="110"/><br>Tibetan</td> </tr> <tr> <td align="center"><img src="./Ethiopian.png" width="110"/><br>Ethiopian</td> <td align="center"><img src="./Adlam.png" width="110"/><br>Adlam</td> <td></td> <td></td> <td></td> <td></td> </tr> </table>

🏷 Label Format

CSV

Parquet

Same structure as CSV but stored in columnar format.

📊 Dataset Statistics

🎯 Intended Use

This dataset is designed for:

• —Multilingual OCR training
• —Vision-language model pretraining
• —Script recognition research
• —Robust text recognition under distortion 🚫 Not intended for bypassing real-world CAPTCHA security systems.

⚙️ Example Loading

import pandas as pd

df = pd.read_csv("labels.csv")
print(df.head())