NWPU-Refer
收藏资源简介:
NWPU-Refer是一个大规模的、多样化的遥感图像分割数据集,包含来自30多个国家的15,003张高分辨率图像(1024-2048px),支持单对象、多对象和非对象分割场景。该数据集的创建旨在促进遥感图像分割领域的发展,解决现有数据集在分辨率、场景多样性和类别覆盖方面的局限性。数据集的应用领域包括遥感图像分割、目标检测、场景理解等,旨在提高遥感图像分割模型的泛化能力和实际应用性。
NWPU-Refer is a large-scale and diverse remote sensing image segmentation dataset. It contains 15,003 high-resolution images ranging from 1024 to 2048 pixels sourced from over 30 countries, and supports single-object, multi-object and non-object segmentation scenarios. This dataset is developed to advance the development of the remote sensing image segmentation field, addressing the limitations of existing datasets in terms of resolution, scene diversity and category coverage. Its application areas include remote sensing image segmentation, object detection, scene understanding and other related fields, aiming to improve the generalization ability and practical applicability of remote sensing image segmentation models.
NWPU-Refer数据集概述
数据集基本信息
- 数据集名称: NWPU-Refer
- 官方仓库: https://github.com/CVer-Yang/NWPU-Refer
- 数据集描述: 大规模遥感图像分割参考数据集与基准
数据集获取
- 下载地址: https://pan.baidu.com/s/1HzLVXSqk_m4zWWJ-EJ5SkA
- 提取码: NWPU
相关资源
- 开源内容: 数据集与代码将在官方仓库中开源




