UIT-HWDB

UIT-HWDB数据集由越南胡志明市信息科技大学开发，旨在解决越南语手写图像识别的挑战。该数据集分为UIT-HWDB-word（110,745张非限制性手写单词图像）和UIT-HWDB-line（7,273张非限制性手写行图像）两部分，通过转移方法构建，确保了手写图像的自然属性和复杂性。数据集的创建过程涉及从在线手写数据集VNOnDB中继承手写字符，并采用颜色变化增强图像自然度。UIT-HWDB数据集适用于评估和改进手写识别技术，特别是在处理越南语这种资源较少的语言时，有助于推动相关研究和技术发展。

The UIT-HWDB dataset was developed by the University of Information Technology, Ho Chi Minh City, Vietnam, to address the challenges of Vietnamese handwritten image recognition. It is divided into two subsets: UIT-HWDB-word, which contains 110,745 unconstrained handwritten word images, and UIT-HWDB-line, which includes 7,273 unconstrained handwritten line images. The dataset was constructed via transfer-based methodologies to preserve the natural attributes and complexity of handwritten images. The dataset creation process involves inheriting handwritten characters from the online handwritten dataset VNOnDB, and adopting color variation techniques to enhance the naturalness of the images. The UIT-HWDB dataset is suitable for evaluating and advancing handwritten recognition technologies, especially when dealing with low-resource languages such as Vietnamese, and it helps promote relevant research and technological development.