AIGeeksGroup/VaseVQA

VASEVQA数据集是一个针对古代希腊/罗马陶器和花瓶的视觉问答数据集，包含31,773张花瓶图像。数据集采用对话式问答对格式（LLaVA格式），分为单视角和多视角两种数据结构。每个花瓶都会被注释关于其材质、技术、形状、来源、年代、归属、装饰描述以及综合信息的问题。The VASEVQA dataset is a Visual Question Answering dataset for ancient Greek/Roman pottery and vases, containing 31,773 vase images. The dataset is in a conversational QA pair format (LLaVA format) and has two data structures: single-view and multi-view. Each vase is annotated with questions regarding its fabric, technique, shape name, provenance/origin, dating information, attribution details, decorative descriptions, and overall comprehensive information.