SECOND-CC
收藏资源简介:
SECOND-CC数据集由伊斯坦布尔技术大学的研究团队创建,专为遥感图像变化描述任务设计。该数据集包含6041对高分辨率RGB图像对、语义分割图以及30205条描述图像差异的句子,涵盖了多种真实场景,如模糊、光照差异、视角变化和图像配准误差等挑战。数据集通过语义分割图和RGB图像的结合,提供了丰富的多模态信息,旨在解决遥感图像变化描述中的复杂问题。该数据集的应用领域包括灾害响应、城市规划、环境监测等,能够为这些领域提供详细的变化描述和及时的行动建议。
The SECOND-CC dataset was created by a research team from Istanbul Technical University, and is specifically designed for the task of remote sensing image change captioning. This dataset includes 6041 pairs of high-resolution RGB images, semantic segmentation maps, and 30205 sentences describing image differences, covering diverse real-world scenarios with challenges such as blurriness, lighting variations, viewpoint changes, and image registration errors. By integrating semantic segmentation maps and RGB images, the dataset offers rich multimodal information, aiming to address complex problems in remote sensing image change captioning. Its application domains cover disaster response, urban planning, environmental monitoring and other fields, and it can provide detailed change descriptions and timely action recommendations for these scenarios.

- 1Robust Change Captioning in Remote Sensing: SECOND-CC Dataset and MModalCC Framework伊斯坦布尔技术大学 · 2025年



