ILIAS (Instance-Level Image retrieval At Scale)
收藏资源简介:
ILIAS是一个用于实例级图像检索的大规模数据集,由捷克技术大学布拉格的电气工程学院创建。该数据集包含1000个对象实例的查询和正样本图像,这些图像是在具有挑战性的条件和多样化领域中手动收集的。为了在不额外注释工作的情况下避免假阴性,数据集仅包含在2014年之后出现的查询对象。ILIAS旨在支持未来的图像到图像和文本到图像检索研究,并作为评估基础视觉和语言模型表示的大规模基准。
ILIAS is a large-scale dataset for instance-level image retrieval, created by the Faculty of Electrical Engineering, Czech Technical University in Prague. This dataset contains query and positive sample images for 1000 object instances, which were manually collected under challenging conditions and across diverse domains. To avoid false negatives without additional annotation work, the dataset only includes query objects that emerged after 2014. ILIAS aims to support future research on image-to-image and text-to-image retrieval, and serve as a large-scale benchmark for evaluating representations of foundational vision and language models.




