DHJ: A Database of Handwritten Jawi for Recognition Research

This paper proposes a database of handwritten Jawi (DHJ) for conducting the research in this area. Previously, a printed Jawi dataset has been proposed and this handwritten dataset is aim to extend the dataset collection in Jawi database. The DHJ database contains handwritten Jawi character, handwritten Jawi word, and handwritten Jawi sentence. This dataset was recorded from ten writers with different age, gender, and educational background. The participants were asked to fill a form, which had three parts: character part, word part, and sentence part. The DHJ was stored in jpeg digital image format and each of image has three different sizes. Totally, we have 3,810 handwritten Jawi characters, 300 handwritten Jawi words, and 40 handwritten Jawi sentences.