City-4Band数据集

City-4Band数据集将每对配准的图像按照可见光图像、短波红外图像、中波红外图像、长波红外图像的顺序按通道拼接存储为tiff文件，按8：1：1的比例将1352个tiff文件划分为训练集、验证集和测试集，分别放在images文件夹的train、val、test三个文件夹下。文件命名例如20241025_142228_0000_1.tiff，20241025为时间信息，142228和0000为序列信息，1为帧数信息。语义分割标签文件为与四波段图像分辨率相同的八位图像文件，保存为png格式，命名和文件夹组织与四波段图像相同且一一对应。The City-4Band dataset stacks each pair of registered images as TIFF files in the order of visible light image, short-wave infrared (SWIR) image, mid-wave infrared (MWIR) image, and long-wave infrared (LWIR) image along the channel dimension. A total of 1352 TIFF files are split into training, validation and test sets with an 8:1:1 ratio, which are placed in the train, val and test subfolders under the images directory respectively. An example of file naming is 20241025_142228_0000_1.tiff, where 20241025 represents the timestamp, 142228 and 0000 are sequence information, and 1 denotes the frame number. Semantic segmentation label files are 8-bit image files with the same resolution as the 4-band images, saved in PNG format, with identical naming and folder organization as the 4-band images and maintain one-to-one correspondence with them.

搜集汇总

背景与挑战

背景概述

City-4Band数据集是一个多波段遥感图像集合，整合了可见光、短波红外、中波红外和长波红外四个波段的图像，按通道拼接存储为tiff格式，并按照8:1:1的比例划分为训练集、验证集和测试集，共包含1352个文件。数据集提供了与图像分辨率相同的语义分割标签，以png格式存储，命名和组织结构与图像文件一一对应，适用于遥感图像分析和语义分割任务。