TLL

该数据集名为TLL，来源于一个知名网站，包含了用户认为相似的图像对，共计6,016对，展示了相似性的多种解读。这些图像对是用户自发上传的，因此在数据集中呈现出有趣的多样性。通过人工实验验证了图像对的配对一致性。该数据集的规模为6,016对图像，其任务是预测人类的感知相似性。

This dataset, designated as TLL, is derived from a prominent online platform. It encompasses 6,016 pairs of images that users consider visually similar, demonstrating diverse understandings of visual similarity. Since these image pairs were voluntarily uploaded by users, the dataset exhibits intriguing inherent diversity. The pairing consistency of these image pairs has been validated via human subject experiments. With a total of 6,016 image pairs, the core task of this dataset is to predict human-perceived visual similarity.