AlexU-Word

AlexU-Word数据集由亚历山大大学工程学院创建，旨在收集涵盖所有阿拉伯字母可能形状的大量孤立阿拉伯单词样本。该数据集当前版本包含25114个样本，来自109个独特的阿拉伯单词，覆盖了所有字母的所有可能形状。数据集的创建过程涉及从907名不同作者收集样本，并经过严格的处理和验证。AlexU-Word数据集主要用于闭词汇的孤立单词识别问题，旨在通过大量样本训练和评估基于分段字母识别的阿拉伯手写识别系统。

The AlexU-Word dataset was developed by the Faculty of Engineering, Alexandria University, with the objective of collecting a large corpus of isolated Arabic word samples that encompass all possible shapes of every Arabic letter. The current version of this dataset contains 25,114 samples derived from 109 unique Arabic words, with full coverage of all potential shapes for each Arabic letter. The dataset construction process involved gathering samples from 907 distinct authors, followed by rigorous processing and validation. The AlexU-Word dataset is primarily applied to closed-vocabulary isolated word recognition tasks, and is designed to train and evaluate Arabic handwriting recognition systems based on segmented letter recognition using a substantial number of samples.