mengcy/LAION-SG

LAION-SG是一个大规模的场景图（SG）数据集，包含高质量的注释，描述了复杂场景中多个对象的属性和关系。数据集主要用于图像到文本的任务，所有注释使用英语。数据集包含540,005个SG-图像对，分为训练、验证和测试集。数据集的创建基于LAION-Aesthetics V2 (6.5+)数据集，使用GPT-4o进行自动化注释。

LAION-SG is a large-scale dataset with high-quality structural annotations of scene graphs (SG), which precisely describe attributes and relationships of multiple objects, effectively representing the semantic structure in complex scenes. The dataset uses English for annotations and contains 540,005 SG-image pairs annotated with objects, attributes, and relationships, split into training, validation, and test sets. The dataset is built on the LAION-Aesthetics V2 (6.5+) dataset, using GPT-4o for automated annotation to ensure comprehensiveness, systematic structure, and precision in the annotations.