Here, we disseminate a new handwritten digits-dataset, termed Kannada-MNIST, for the Kannada script, that can potentially serve as a direct drop-in replacement for the original MNIST dataset.

Data Collection

This dataset is based off of the efforts of 65 volunteers from Bangalore, India, who are native speakers and users of the Kannada language and the script. This was curated to serve as a direct one-to-one drop-in replacement for the original MNIST dataset (akin to Fashion-MNIST and K-MNIST datasets).

65 volunteers were recruited in Bangalore, India, who were native speakers of the language as well as day-to-day users of the numeral script. Each volunteer filled out an A3 sheet containing a 32 ¡Á 40 grid. This yielded filled-out A3 sheets containing 128 instances of each number which we posit is large enough to capture most of the natural intra-volunteer variations of the glyph shapes. All of the sheets thus collected were scanned at 600 dots-per-inch resolution using the Konica Accurio-Press-C6085 scanner that yielded 65 4963 ¡Á 3509 png images.