Bangla Handwritten Dolil Dataset (653 Images)

The sample size is 653 images of Bangla Handwritten Dolil (land records) and it is generated as part of a sparse solution to archivelomg data. Dolil is the official documents land related which reflect ownership, transfers and agreements. Such documents have historic significance in Bangladesh in a legal context and in the context of cultural heritage. Because of the differences in hand writing circles, ink, and paper conditions, processing of Bangla handwritten land records would result in great challenges in optical character recognition (OCR), hand writing recognition, and digitalizing.

The data is prepared to guide studies towards Bangla OCR development, document image processing, handwriting recognition and digitization of ancient land records. The photo images are performed with high-resolution in order to maintain the handwriting and paper characteristics.

The dataset of Bangla handwritten land records is one of the earliest accessible form of Bangla handwritten records it will form a unique resource to researchers of the field of computer vision, pattern recognition, natural language processing (NLP) and digital humanities. It is hoped that it would be used to help with projects that involve document digitization, land records management and preservation of Bangladesh cultural and the legal heritage.

本数据集共包含653张孟加拉语手写Dolil（土地登记文书）图像，其构建是为解决档案数据稀疏性问题而开展的一项探索性工作。Dolil为孟加拉语中与土地相关的官方文书，用于记录土地所有权、流转及相关协议。此类文书在孟加拉国的法律体系与文化遗产语境中均具备重要历史价值。由于书写笔迹风格、墨水材质及纸张保存状态存在差异，孟加拉语手写土地登记文书在光学字符识别（Optical Character Recognition，OCR）、手写识别及数字化处理过程中均面临极大挑战。

本数据集的构建旨在为孟加拉语OCR开发、文档图像处理、手写识别及古代土地登记文书数字化相关研究提供支撑。为完整保留手写笔迹与纸张的原始特征，所有图像均采用高分辨率拍摄。

本孟加拉语手写土地登记文书数据集是目前可获取的最早一批孟加拉语手写文书资源之一，将为计算机视觉、模式识别、自然语言处理（Natural Language Processing，NLP）及数字人文领域的研究者提供独特的研究资源。本数据集有望应用于文档数字化、土地登记文书管理以及孟加拉国文化与法律遗产保护相关项目，助力相关研究与实践工作。