Omarrran/40K_kashmiri_text_and_image_dataset

40K Kashmiri Words with images

Kashmiri (words) Image and Text Dataset for OCR Models

This repository contains a dataset specifically curated for training and testing Optical Character Recognition (OCR) models on Kashmiri language text. The dataset includes a large collection of images with their corresponding labels in a CSV file, designed to aid in the development of robust OCR solutions for the Kashmiri script.

---

Directory Structure:

Zip File/
├── images/
│   ├── 000001.png
│   ├── 000002.png
│   ├── 000003.png
│   ├── ...
│   ├── 031000.png
├── labels.csv

Description:

Zip File/: The root directory containing all the dataset files and subfolders.

images/: A folder housing 40,799 image files, sequentially named from 000001.png to 040799.png. Each image contains text written in the Kashmiri language.

labels.csv: A CSV file mapping each image file to its corresponding text label. The file format includes two columns:

Sample fromat

Image ID	Text	Image File
000001	لیونڈ
000002	لایم
000003	ٹیرف
000004	تلاان
000005	آرکیالوجیکل
000006	چتوڈکہ
000007	پریانشو
000008	کالیسٹریٹسنک
000009	جولاے

Usage

This dataset is suitable for:

Training OCR Models: Develop OCR systems for the Kashmiri script.

Testing OCR Algorithms: Benchmark and evaluate OCR performance on a diverse set of Kashmiri text images.

Research: Conduct linguistic or technical research on OCR in low-resource languages.

Example Usage in Python

1

Extract the Zip file then ;

2

import pandas as pd
from PIL import Image

# Load labels
labels = pd.read_csv('labels.csv')

# Example: Display an image and its corresponding label
image_path = 'images/000001.png'
image = Image.open(image_path)
print(f"Label: {labels[labels['image_name'] == '000001.png']['text_label'].values[0]}")
image.show()

Dataset Details

Number of Images: 40,799

Language: Kashmiri

File Format:

Images: .png

Labels: .csv

---

Contribution

If you encounter any issues or have suggestions for improving this dataset, feel free to open an issue or submit a pull request.

---

License

This dataset is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0). Please provide appropriate credit when using this dataset in your projects.

Citation:

@misc{dataset2024,
  title        = {Omarrran/40K_kashmiri_text_and_image_dataset},
  author       = {HAQ NAWAZ MALIK},
  year         = {2024},
  url          = {https://huggingface.co/datasets/Omarrran/40K_kashmiri_text_and_image_dataset/},
  note         = {Contains 40,799 images with labels }
}