Image descriptions for 7471 of the DEArt images, obtained via the Zooniverse crowdsourcing platform

This dataset contains crowdsourced image descriptions for 7471 images from the DEArt dataset. There are typically 4-5 descriptions per image, provided by volunteers of the Zooniverse crowdsourcing platform. The guidelines we used, together with some examples of possible captions, are also published in Zooniverse. We are in debt with all the Zooniverse volunteers and coordinators, and in particular with Samantha Blickhan, for making this possible. Some image descriptions were filtered out whenever they did not contain any textual data, nevertheless the dataset may still contain some descriptions that are not according to the guidelines we provided to the annotators. The dataset is a data export from Zooniverse in CSV format, in which the user information has been anonymized. The CSV also includes a field with the filename of the described image in the DEArt dataset.本数据集包含来自DEArt数据集的7471张图像的众包图像描述。每张图像通常配有4至5条描述，均由Zooniverse众包平台的志愿者撰写。本次标注所使用的标注指南及若干示例标注文本，亦已在Zooniverse平台发布。我们谨向所有Zooniverse的志愿者与协调人员，尤其是Samantha Blickhan，致以诚挚谢意，感谢他们促成本数据集的完成。部分未包含有效文本内容的图像描述已被筛除，但本数据集仍可能存在少量不符合我们向标注人员提供的指南要求的描述。本数据集为从Zooniverse导出的CSV格式数据，其中用户信息已完成匿名化处理。该CSV文件还包含一列字段，对应DEArt数据集中被标注图像的文件名。