kalixlouiis/MyanmarOCR-ImageText

---
dataset_info:
features:
- name: image
dtype: image
- name: text
dtype: string
- name: style
dtype: string
splits:
- name: train
num_bytes: 532905094.304
num_examples: 41664
download_size: 266331659
dataset_size: 532905094.304
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
language:
- my
size_categories:
- 10K<n<100K
license: cc-by-4.0
task_categories:
- feature-extraction
tags:
- ocr
- image-to-text
pretty_name: MyanmarOCR-ImageText
---
# 🇲🇲 MyanmarOCR-ImageText Dataset

A clean and diverse **Burmese Image-to-Text** dataset for OCR and multimodal AI research.

---

## 📌 Summary

- **Total images:** 41,664
- **Unique Burmese text entries:** 1,139
- **Styles per text:** 32 variations each
- **Resolution:** 512 × 512
- **File types:** PNG/JPG images
- **Dataset split:** train only
- **Use cases:** OCR, I2T (image-to-text), VLM pretrain/fine-tune

All text is **Burmese only**.
No English words and no punctuation like: `? , ' " -`

---

## 🔡 Text Content

Includes:
- Common words & Pali words
- Signs and short phrases
- Full Myanmar Unicode support

> Variety of Myanmar spellings and writings are included
> (မြန်မာအက္ခရာတွေနဲ့ စကားလုံးမျိုးစုံပါဝင်ပါတယ်)

---

## 🎨 Style Variations

Each text appears in **32 visual styles** with differences in:
- font
- color
- texture
- rotation (small angle)
- background patterns

This helps models generalize across real-world environments.

---

## 🧩 Data Format

Each sample includes:

| Column | Type | Description |
|--------|------|-------------|
| `image` | image | 512×512 Burmese rendered text |
| `text` | string | Ground truth Burmese label |
| `style` | string | Style ID (e.g., `style_01`) |

Example record:

```json
{
"image": "<image>",
"text": "မြန်မာနိုင်ငံ",
"style": "style_07"
}
```
---

## 🧪 Usage

```python
from datasets import load_dataset

ds = load_dataset("kalixlouiis/MyanmarOCR-ImageText", split="train")
print(ds[0])

ds[0]["image"].show()
```

---

## 🎯 Intended Purposes

- Burmese OCR training
- Scene text model finetuning
- Vision-language pretraining
- Synthetic-to-real text recognition research
- Benchmark for Myanmar multimodal AI

---

## ⚠️ Limitations

- Synthetic images only — not real photos/signboards
- No English text or punctuation
- No complex layout structures (single word/short text per image)

---

## 📄 License

This dataset is released under the **Creative Commons Attribution 4.0 International (CC-BY-4.0)** license.

You are free to:
- ✔ Share — copy and redistribute for any purpose
- ✔ Adapt — modify, transform, and build upon the data

As long as you:
- **Give appropriate credit**
- **Indicate changes**
- **Provide a link to the license**

📌 License Text: https://creativecommons.org/licenses/by/4.0/

---

## ✨ Acknowledgment

Created by **[@kalixlouiis](https://huggingface.co/kalixlouiis)**
with the goal of improving **Myanmar OCR and AI research**.

If you use this dataset in your research or applications,
please cite and provide a link to the dataset page on Hugging Face:

🔗 https://huggingface.co/datasets/kalixlouiis/MyanmarOCR-ImageText

---

## 📚 Citation

```bibtex
@dataset{kalixlouiis2025myanmarocr,
title = {MyanmarOCR-ImageText},
author = {Kalix Louis},
year = {2025},
publisher = {Hugging Face},
howpublished = {\url{https://huggingface.co/datasets/kalixlouiis/MyanmarOCR-ImageText}},
license = {CC-BY-4.0}
}
```
---