Hyperspectral City V1.0
收藏资源简介:
Hyperspectral City V1.0数据集由南京大学创建,专注于城市自动驾驶场景的语义分割。该数据集采用新开发的 hyperspectral 相机,旨在解决现有数据集视觉质量不足的问题。数据集包含367帧用于训练的粗标注 hyperspectral 图像和55帧用于测试的精细标注图像。创建过程中,数据在上海市多种环境下收集,包括交通繁忙区域、著名建筑、中央商务区等,涵盖晴天和阴天,以及白天、夜晚和日落时的光照条件。该数据集适用于自动驾驶领域的场景理解和语义分割任务,旨在提高自动驾驶系统在复杂环境下的识别和决策能力。
Hyperspectral City V1.0 is a dataset created by Nanjing University, focusing on semantic segmentation tasks for urban autonomous driving scenarios. It adopts a newly developed hyperspectral camera to address the issue of insufficient visual quality in existing datasets. The dataset includes 367 frames of coarsely annotated hyperspectral images for model training and 55 frames of finely annotated images for testing. During its development, the data was collected across various environments in Shanghai, including heavy traffic areas, famous landmarks, central business districts, etc., covering both sunny and cloudy weather, as well as lighting conditions during daytime, nighttime and sunset hours. This dataset is applicable to scene understanding and semantic segmentation tasks in the autonomous driving domain, aiming to improve the recognition and decision-making capabilities of autonomous driving systems in complex environments.




