Handwritten Notations of Digits and Fractions in Telugu Script : A Language of India

This repository consists of three folders of Handwritten Telugu Fraction numerals. The folder "preprocessed_img" is a balanced dataset of a total of 4000 Telugu Fraction notations (32x32 4000 images), each having 500 images per class (per digit, i.e., 8 classes). The 8 classes are ౸ (haḷḷi), ౹ (k̄alu), ౺ (ara), ౻ (mukk̄alu), ౦ (sunna), ౼ (v̄isamu), ౽ (paraka) and ౾ (muvv̄isamu). While the datasets for handwritten "0 to 9" digit images of Telugu and many other Indian Scripts are prevalent, this is the first-ever initiative to curate a Telugu Fractions-only dataset of handwritten single-digit images. A total of 80 people contributed to this dataset with their handwriting. Those handwritten sets can be found in the folder "pages_grey". The preprocessed images from this dataset helped hugely in researching handwritten digit recognition systems for Telugu Script. Telugu is a largely spoken Dravidian Language from the southern part of India. If you land upon this dataset and want to work on the same or anything better, starting from starch with the non-processed images in the folder "digit_cropped" is suggested. Please cite this creator if you adapt this dataset to your work.

How to Cite? Check below -

Vempati, Lakshmi Sravani (2024), “Handwritten Notations of Digits and Fractions in Telugu Script : A Language of India”, Mendeley Data, V1, doi: 10.17632/mpjbr7nfv9.1