Fields of The World (FTW)
收藏资源简介:
Fields of The World (FTW) 是一个用于全球农业田地边界分割的机器学习基准数据集,涵盖了24个国家的农业田地数据。该数据集包含70,462个样本,每个样本都包含实例和语义分割掩码,并与多日期、多光谱的Sentinel-2卫星图像配对。FTW的创建旨在解决当前机器学习方法在田地实例分割中地理覆盖不足、准确性低和泛化能力差的问题。数据集的应用领域包括农业监测、可持续农业和气候变化政策等,旨在通过自动化的田地边界提取来支持全球农业数据的获取和分析。
Fields of The World (FTW) is a machine learning benchmark dataset for global agricultural field boundary segmentation, covering agricultural field data from 24 countries. The dataset contains 70,462 samples, each with instance and semantic segmentation masks paired with multi-temporal, multi-spectral Sentinel-2 satellite imagery. Developed to address the shortcomings of current machine learning methods in field instance segmentation—including insufficient geographic coverage, low accuracy and poor generalization ability—FTW has application domains such as agricultural monitoring, sustainable agriculture and climate change policy, with the core goal of supporting the acquisition and analysis of global agricultural data through automated field boundary extraction.

- 1Fields of The World: A Machine Learning Benchmark Dataset For Global Agricultural Field Boundary Segmentation亚利桑那州立大学 · 2024年



