moymarksama/CUHK-PEDES

CUHK-PEDES是一个数据集，包含图像和文本对，主要用于行人描述和检索任务。它由香港中文大学开发，包含238,768个训练示例，每个示例由一张行人图像和相关的文本描述组成。该数据集常用于计算机视觉和自然语言处理的交叉研究，如图像-文本匹配、行人重识别和跨模态检索。

CUHK-PEDES is a dataset consisting of image-text pairs, primarily used for pedestrian description and retrieval tasks. Developed by The Chinese University of Hong Kong, it includes 238,768 training examples, each comprising a pedestrian image and an associated text description. The dataset is commonly applied in cross-disciplinary research in computer vision and natural language processing, such as image-text matching, person re-identification, and cross-modal retrieval.