Groudtruth Images for testing character segmentation algorithms from palm leaf images

This dataset consists of manually segmented grantha characters from palmleaflets.Grantha characters was the script used in southern India to write sanskrit language.This dataset will serve as ground truth for testing segmentation algorithms for extraction of characters from palm leaf images.The palm leaflets are taken from the manuscript "Isadi Upanishad" obtained from Oriental Research Institute under University of Kerala,India.It consists of two input palm leaflets and the corresponding cropped characters of each line arranged in different folders.

本数据集由人工分割自贝叶经上的格兰塔文字，格兰塔文字是南印度用以书写梵文之文字。该数据集将作为测试字符提取算法从贝叶经图像中提取字符的基准。贝叶经取自印度喀拉拉邦大学东方研究所收藏的《伊莎迪奥义书》。数据集包含两份输入贝叶经以及每行对应的裁剪字符，并按不同文件夹排列。