CGCL

CGCL数据集由乌克兰国立技术大学“伊戈尔·西科尔斯基基辅理工学院”创建，包含超过4000张来自基辅圣索菲亚大教堂的历史雕刻字母（XI-XVIII世纪）的图像，涵盖34种字母类型。该数据集通过图像预处理，为机器学习方法如多类别逻辑回归和2D卷积神经网络提供了训练和预测的素材。创建过程中，数据集不断通过新鲜贡献进行扩充。应用领域主要集中在通过机器学习技术自动识别历史雕刻字母，解决雕刻字母因复杂几何形状、难以辨识的形态和低统计代表性带来的识别难题。

The CGCL dataset was created by the National Technical University of Ukraine "Igor Sikorsky Kyiv Polytechnic Institute". It contains over 4,000 images of historical carved letters (11th–18th centuries) from Saint Sophia Cathedral in Kyiv, covering 34 letter types. This dataset has undergone image preprocessing to provide training and prediction resources for machine learning methods such as multi-class logistic regression and 2D convolutional neural networks. Throughout its curation process, the dataset has been continuously augmented with new contributions. Its primary application focuses on the automatic recognition of historical carved letters via machine learning techniques, addressing the recognition challenges stemming from the complex geometric shapes, ambiguous morphologies, and low statistical representativeness of such carved letters.