Marathi Handwritten Text Images for OCR Development

2520 handwritten text images were curated from native Marathi speakers to ensure variety in handwriting styles and character variations. The dataset supports Optical Character Recognition (OCR) systems, handwriting analysis, and language research. It was created by Process-Venue and last updated in February 2025.

Use Cases

• Train OCR models to recognize handwritten Marathi characters from the image data.
• Analyze handwriting style variations across native speakers using the curated image samples.
• Benchmark the performance of Devanagari script recognition algorithms on 2520 handwritten images.
• Develop data augmentation techniques for handwritten text by leveraging the variety of character variations present.

Strengths

• Contains 2520 handwritten text images.
• Curated from native Marathi speakers to ensure authentic handwriting styles.

Limitations

• Sample size of 2520 images may be insufficient for training large, modern deep learning models.
• Unknown license restricts clarity on permissible commercial or research use.
• No column or label information provided, limiting structured analysis.

Provenance

Source

Process-Venue via Hugging Face.

Collection Method

Curated from handwritten text provided by native Marathi speakers.

Freshness

Last updated on 2025-02-20.