LRS-VQA

该数据集是一个包含7,333个问答对的综合基准，覆盖了8个类别，其中的图像最长可达27,328像素。数据集分为三个部分：LRS-FAIR、LRS-Bridge和LRS-STAR，分别包含2,272、1,062和3,999个问答对。该数据集的规模为7,333个问答对，其任务重点在于视觉-语言理解和问答。

This dataset is a comprehensive benchmark comprising 7,333 question-answer pairs, covering 8 categories, with images up to 27,328 pixels in length. The dataset is divided into three subsets: LRS-FAIR, LRS-Bridge, and LRS-STAR, which contain 2,272, 1,062, and 3,999 question-answer pairs respectively. With a total of 7,333 question-answer pairs, this dataset focuses on vision-language understanding and question answering tasks.

搜集汇总

背景与挑战

背景概述

LRS-VQA是一个专注于大型遥感图像感知的基准测试数据集，包含1,657张高分辨率图像和7,333个问答对，覆盖8种问题类型，旨在评估大型视觉语言模型在高分辨率遥感图像上的感知能力。