BanglaLekha-Isolated

BanglaLekha-Isolated是一个全面的孟加拉手写文字数据集，由大学自由艺术孟加拉国创建。该数据集包含84种不同的孟加拉手写数字、基本字符和复合字符，共计166,105个样本，来源于孟加拉国多个地理区域的各年龄段人群。数据集创建过程中，通过提供特定格式的表格收集样本，并进行了预处理，如背景反转、噪声去除和图像调整。该数据集主要用于孟加拉手写文字识别研究，同时也可用于性别、年龄、地区等特征的自动识别，以及手写质量评估和反馈方法的研究。

BanglaLekha-Isolated is a comprehensive Bengali handwritten character dataset developed by the University of Liberal Arts Bangladesh. It includes 84 distinct categories of Bengali handwritten digits, basic characters and compound characters, with a total of 166,105 samples collected from individuals across all age groups and multiple geographic regions in Bangladesh. During the dataset development process, samples were collected via specially formatted forms, followed by preprocessing steps such as background inversion, noise removal and image adjustment. This dataset is primarily used for research on Bengali handwritten character recognition, and can also be applied to studies on automatic recognition of attributes such as gender, age and region, as well as research on handwriting quality assessment and feedback methods.