GeneCIS

该数据集是一个以关键词为条件的图像相似度测量基准，包含了四个子任务，主要关注于改变给定图像中的属性或对象，或者是对其进行聚焦。在此基准下，模型需要根据给定关键词找到与查询图像最相似的一组图片。平均而言，每个查询条件对应大约13.8张图像。这一任务被称为关键词条件下的图像相似度测量。

This dataset is a keyword-conditioned image similarity measurement benchmark consisting of four subtasks. These subtasks primarily focus on modifying attributes or objects within given images, or performing focused adjustments. Under this benchmark, models are required to retrieve the set of images most similar to a query image based on a specified keyword. On average, each query condition corresponds to approximately 13.8 images. This task is termed keyword-conditioned image similarity measurement.