SALAMI

SALAMI数据集是由维也纳技术大学计算机视觉实验室创建的，包含250张古代手稿图像，用于评估图像中文本的清晰度。数据集中的图像来自48种不同的手稿和8种语言家族，每张图像都有对应的清晰度和不确定性空间图。创建过程中，研究人员采用了半自动随机选择和人工修订的方法，确保数据集的多样性和代表性。该数据集主要应用于开发和验证客观的清晰度评估方法，旨在解决数字文本修复领域缺乏量化评估标准的问题。

The SALAMI dataset was created by the Computer Vision Laboratory of Vienna University of Technology. It contains 250 ancient manuscript images for evaluating the legibility of text within images. The images in the dataset are sourced from 48 distinct manuscripts spanning 8 language families, and each image is paired with a corresponding spatial map of legibility and uncertainty. During its creation, researchers adopted a semi-automatic random selection combined with manual revision approach to ensure the diversity and representativeness of the dataset. This dataset is primarily used for developing and validating objective legibility assessment methods, aiming to address the lack of quantitative evaluation criteria in the field of digital text restoration.