ARVSU

该数据集名为ARVSU，包含了大量视觉场景图像的变体，每个场景都附有标注的语句及其对应的交谈对象。这一设计旨在模拟在多种社交环境中，人与智能系统之间的互动。该数据集是通过亚马逊土耳其机器人（Amazon Mechanical Turk）创建的，其中包含了指向场景中不同实体的语句，包括摄影师、视线中的人以及其他对象。该数据集的任务是对视觉场景中的交谈对象进行识别。

This dataset, named ARVSU, contains numerous variants of visual scene images, each paired with annotated utterances and their corresponding interlocutors. It is designed to simulate interactions between humans and intelligent systems across diverse social environments. Constructed via Amazon Mechanical Turk, this dataset includes utterances that refer to different entities within the scenes, such as photographers, people within the line of sight, and other objects. The task associated with this dataset is to identify the interlocutors in visual scenes.