CityEQA-EC
收藏资源简介:
CityEQA-EC数据集是首个面向城市空间的具体问题回答任务的开源数据集,由国防科技大学和清华大学共同研发。该数据集基于EmbodiedCity平台,包含六个类别共1412个人工标注的任务实例,这些任务实例涵盖了丰富的城市地标和空间关系。数据集的构建过程包括原始问答生成和任务补充两个步骤,最终通过人工审核确保数据质量。该数据集的应用领域主要针对城市空间智能,旨在促进Embodied Question Answering在城市环境中的应用和研究。
CityEQA-EC is the first open-source dataset dedicated to urban spatial-specific question answering tasks, co-developed by the National University of Defense Technology and Tsinghua University. Built upon the EmbodiedCity platform, this dataset comprises 1412 manually annotated task instances across six categories, covering a wide range of urban landmarks and spatial relationships. The construction of this dataset involves two main stages: raw question-answer generation and task supplementation, with final manual review conducted to ensure high data quality. Its application scenarios primarily focus on urban spatial intelligence, aiming to promote the application and research of Embodied Question Answering in urban environments.




