The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
Sindhi Character Recognition Dataset
Overview
This is a custom-made image dataset built for Optical Character Recognition (OCR) on the Sindhi language. It covers the first 25 letters of the Sindhi alphabet, with 50 images collected for each letter, organized into labeled folders so the data is ready to use for training and testing character recognition models.
Dataset Details
- Category: OCR
- Source: Custom made
- Language: Sindhi
- Classes: 25 (first 25 letters of the Sindhi alphabet)
- Images per class: 50
- Total images: 1,250
- Diversity: Images include augmentation (rotation, scaling, noise, etc.) and different perspectives/angles, so the model sees each letter written and captured in more than one way.
Folder Structure
Each letter has its own folder, and all 50 images for that letter live inside it. This makes the dataset ready to load directly with common ML libraries (e.g. ImageFolder in PyTorch or flow_from_directory in Keras) without extra reformatting.
Sindhi_Character_Recognition_Dataset/
├── 01_alphabet/
│ ├── img_001.jpg
│ ├── img_002.jpg
│ └── ... (50 images)
├── 02_alphabet/
│ ├── img_001.jpg
│ └── ... (50 images)
├── ...
└── 25_alphabet/
└── ... (50 images)
(Replace folder names above with the actual Sindhi letter names/labels used in your data.)
Data Collection & Augmentation
Images were custom collected and then expanded using augmentation techniques and varied perspectives, to help models generalize better instead of overfitting to a single writing style or camera angle.
Intended Use
- Training and evaluating OCR / character recognition models for Sindhi script
- Benchmarking low-resource language OCR systems
- Academic and research use in Sindhi language technology
Limitations
- Only covers the first 25 letters of the Sindhi alphabet (not the full character set)
- 50 images per class is a relatively small sample size, so results may not generalize to large-scale, real-world documents without further data collection
License & Usage
Add your intended license here (e.g. CC-BY-4.0, MIT, or "available on request, non-commercial use only").
Citation
If you use this dataset, please cite it as:
Sindhi Character Recognition Dataset. Sana , 2026.
Contact
- Downloads last month
- 11