helena-balabin/vg_coco_overlap

该数据集包含图像及其相关文本描述的特征信息，主要用于图像与文本的关联分析。数据集的特征包括文件路径、句子ID、文件名、图像ID、原始句子、图像URL、图像宽高比、句子长度、解析树深度、动词数量、图像复杂度评分、平均节点度、平均节点连接性、平均聚类系数、密度、场景图深度、过滤后的场景图深度、连接组件数量、对象数量、关系数量、过滤后的关系数量、视觉基因组图像ID和COCO图像ID等。数据集仅包含一个训练集分割，包含236,184个样本，总大小为79,130,978字节。

This dataset contains feature information of images and their associated textual descriptions, primarily used for image-text association analysis. The features of the dataset include filepath, sentence IDs, filename, image ID, raw sentences, image URL, aspect ratio, sentence length, parse tree depth, number of verbs, image complexity score, average node degree, average node connectivity, average clustering coefficient, density, scene graph depth, filtered scene graph depth, number of connected components, number of objects, number of relations, number of filtered relations, Visual Genome image ID, and COCO image ID. The dataset includes only a training split with 236,184 samples and a total size of 79,130,978 bytes.