BrazilDAM

BrazilDAM是一个基于Sentinel-2和Landsat-8卫星图像的公开数据集，覆盖了巴西国家矿业机构(ANM)列出的所有尾矿坝。该数据集包含2016年至2019年间记录的769个坝址的经过处理的时序图像，旨在用于机器学习基准测试。数据集不仅包含坝址坐标，还包含主要矿石、建设方法、风险类别和潜在损害信息。BrazilDAM特别适用于深度学习方法，如卷积神经网络(CNN)，用于尾矿坝的分类和风险评估，有助于监管机构和当地社区进行风险管理和监测。

BrazilDAM is a public dataset based on Sentinel-2 and Landsat-8 satellite imagery, covering all tailings dams listed by the Brazilian National Mining Agency (ANM). It contains processed time-series images of 769 dam sites recorded between 2016 and 2019, and is designed for machine learning benchmarking. In addition to the coordinates of these dam sites, the dataset also provides information on major ores, construction methods, risk categories and potential damages. BrazilDAM is particularly suitable for deep learning methods such as Convolutional Neural Networks (CNNs) for tailings dam classification and risk assessment, which helps regulatory bodies and local communities conduct risk management and monitoring.

搜集汇总

背景与挑战

背景概述

BrazilDAM是一个基于Sentinel-2和Landsat-8卫星图像的公开数据集，覆盖巴西所有尾矿坝，包含2016年至2019年769个坝址的时序图像，并附带坐标、矿石类型和风险信息。该数据集专为机器学习基准测试设计，特别适用于深度学习方法进行尾矿坝分类和风险评估，以支持监管和社区风险监测。