HP Labs India Indic Handwriting Datasets

One of the core objectives of the Pen-based Applications and Handwriting Recognition Group at HP Labs India was the development of annotated datasets of handwriting in the languages and scripts of developing nations, which are generally conspicuous by their absence.

Annotated handwriting datasets serve multiple purposes. They foster research in handwriting recognition by overcoming an important barrier to research, enable analysis of writing styles to devise handwriting recognition strategies, enable training of robust handwriting recognition algorithms, allow objective evaluation of handwriting recognition algorithms, and can support other applications including script and writer identification.

The HP Labs India Indic Handwriting Datasets were collected using the Dataset Tools available from this site. More information about the dataset creation procedures may be found in our publications.

Thanks to the MILE Lab at Indian Institute of Science, Bangalore, for hosting the datasets for download.

Datasets Available for Download

IWFHR 2006 Online Tamil Handwritten Character Recognition Competition

The Isolated Handwritten Tamil Character Dataset hpl-tamil-iso-char was used for the Tamil Handwritten Character Recognition Competition organized in the context of the 10th International Workshop on Frontiers in Handwriting Recognition (IWFHR-10), La Baule, France, Oct 23-26, 2006. For more details, visit the competition site.