Javanese Character Sewaka Script (JVCSS)

Javanese Character Sewaka Script (JVCSS) is a dataset of handwritten Javanese character images produced from Serat Sewaka Manuscript formatted in MNIST-like format. This dataset is a subset version of object detection dataset - Handwritten Javanese Character Sewaka Detection (HJCS_DETC) dataset. The image are extracted from the main image into single image and pre-processed with grayscaling, blurring, thresholding, resizing and padding to produced uniformly image into 32 x 32 pixels. It contains 20 characters of basic Javanese characters (wyanjana) and have different number of image in each characters varied by the number of characters that appeared in the real document. It may have imbalanced data distribution among the classes. The dataset contains training and test set separated by the source of the image documents.