RSTeller

该数据集包含了一百万多张多模态遥感图像，每张图像都配有利用大型语言模型生成的多个描述性标题。该数据集的目标是民主化高质量标注数据的使用，以便于遥感研究和应用领域的广泛访问。规模上，数据集拥有超过一百万张遥感图像。其任务涵盖了视觉语言建模以及遥感场景理解。

This dataset includes over one million multimodal remote sensing images, with each image accompanied by multiple descriptive captions generated by large language models (LLMs). The core goal of this dataset is to democratize the use of high-quality annotated data, facilitating broad access for remote sensing research and application communities. In terms of scale, this dataset features over one million remote sensing images. Its supported tasks span visual-language modeling and remote sensing scene understanding.