AiR-D

AiR-D是由明尼苏达大学计算机科学与工程系的研究团队创建的一个用于视觉问答（VQA）任务的眼动追踪数据集。该数据集包含1828个场景，旨在支持对人类视觉注意力的理解和机器注意力的研究。数据集通过收集人类在执行视觉推理任务时的眼动数据和答案正确性，为研究机器注意力机制提供了基准。AiR-D数据集的应用领域包括提高视觉问答模型的解释性和性能，以及研究人类和机器在视觉任务中的注意力分配差异。

AiR-D is an eye-tracking dataset for visual question answering (VQA) tasks, developed by a research team from the Department of Computer Science and Engineering at the University of Minnesota. This dataset includes 1,828 scenes and is designed to support research on both the understanding of human visual attention and machine attention. It provides a benchmark for studies on machine attention mechanisms by collecting eye-tracking data and answer accuracy of human participants while they perform visual reasoning tasks. The application scenarios of the AiR-D dataset include improving the interpretability and performance of visual question answering models, as well as investigating the differences in attention allocation between humans and machines in visual tasks.