hr99/graph-based-captions_GBC10M

该数据集包含图像及其相关描述信息，主要特征包括图像的顶点信息（vertices）、图像URL（img_url）、原始描述（original_caption）、简短描述（short_caption）和详细描述（detail_caption）。顶点信息进一步包含边界框（bbox）、描述（descs）、入边（in_edges）、出边（out_edges）和顶点ID（vertex_id）等子字段。数据集分为训练集，包含10,128,574个示例，总大小为75,609,754,045字节。

This dataset contains images and their related descriptions, with main features including image vertices (vertices), image URL (img_url), original caption (original_caption), short caption (short_caption), and detailed caption (detail_caption). The vertices feature further includes subfields such as bounding box (bbox), descriptions (descs), incoming edges (in_edges), outgoing edges (out_edges), and vertex ID (vertex_id). The dataset is divided into a training set, containing 10,128,574 examples with a total size of 75,609,754,045 bytes.