MLRSNet

MLRSNet 是一个用于语义场景理解的多标签高空间分辨率遥感数据集。它提供了从卫星捕获的世界的不同视角。也就是说，它由高空间分辨率的光学卫星图像组成。 MLRSNet 包含 109,161 张遥感图像，被标注为 46 个类别，一个类别中的样本图像数量从 1,500 到 3,000 不等。图像具有 256×256 像素的固定大小，具有各种像素分辨率（~10m 到 0.1m）。此外，数据集中的每张图像都被标记了 60 个预定义类标签中的几个，与每张图像相关的标签数量从 1 到 13 不等。该数据集可用于基于多标签的图像分类、基于多标签的图像检索和图像分割。

MLRSNet is a multi-label high spatial resolution remote sensing dataset for semantic scene understanding. It presents distinct perspectives of the world captured by satellites, namely, it consists of high spatial resolution optical satellite imagery. MLRSNet contains 109,161 remote sensing images annotated with 46 categories, and the number of sample images per category ranges from 1,500 to 3,000. All images have a fixed size of 256×256 pixels, with varying pixel resolutions (approximately 10m to 0.1m). Additionally, each image in the dataset is labeled with several of the 60 predefined class labels, and the number of labels associated with each image varies from 1 to 13. This dataset can be applied to multi-label-based image classification, multi-label-based image retrieval and image segmentation.

搜集汇总

背景与挑战

背景概述

MLRSNet是一个公开的图像分类数据集，规模为1.3GB，包含20.3k个文件，受到广泛关注。