|
所在平台: Udemy |
课程主页: https://www.udemy.com/course/land-use-mapping-utilizing-advanced-machine-learning-models/
课程评论：没有评论
**课程总结：使用开源软件和云平台进行土地覆盖制图的机器学习算法开发** 本课程专注于利用开源软件（QGIS）和Google Earth Engine (GEE)、Google Colab等云计算平台进行土地覆盖制图。课程涵盖了从卫星数据收集、预处理到高级机器学习算法的应用以及模型验证的全过程。 **核心内容：** * **遥感基础与数据采集：** 学习遥感的概念，并通过GEE平台收集和导出卫星影像至Google Drive。 * **数据准备与预处理：** 在QGIS中创建参考/地面真值数据（矢量格式），并将其转换为栅格格式。 * **机器学习算法应用：** 在Google Colab环境中，学习读取和导入栅格及参考数据，并开发和应用决策树 (Decision Trees)、随机森林 (Random Forest) 和 Extra Trees 等高级机器学习算法进行土地覆盖制图。 * **模型验证与特征重要性：** 学习如何验证开发的模型，并探索基于决策树算法的特征重要性建模。 * **结果输出与制图：** 将分类地图导出至本地硬盘，并在QGIS中进行地图布局制作。 **课程亮点：** * **高效的云平台应用：** 强调使用Google Colab，大大减少了环境配置和库安装的问题，提高了分类图生成的速度和可靠性。 * **先进的机器学习技术：** 深入讲解了多种强大的树基机器学习算法，并应用于实际的土地覆盖制图任务。 * **端到端的解决方案：** 提供从数据获取到最终地图发布的完整工作流程。 * **实用的技能培养：** 课程内容对于需要提取和制图特定地物（如高分辨率卫星影像中的城市区域提取）的研究和应用都具有很高的借鉴价值。 总而言之，本课程为学习者提供了一个利用现代遥感和GIS技术，结合先进机器学习方法进行高效、准确土地覆盖制图的全面指导。
Land cover mapping is a critical aspect of Earth's surface monitoring and mapping. In this course, land cover mapping using open-source software (i.e, QGIS) and cloud-computing platforms of the Google Earth Engine (GEE) and Google Colab is covered. The discussed and developed methods can be utilized for different object/feature extraction and mapping (i.e., urban region extraction from high-resolution satellite imagery). Remote sensing is a powerful tool that can be used to identify and classify different land types, assess vegetation conditions, and estimate environmental changes. In this course, you will learn how to collect your satellite data and export it into your hard drive/Google Drive using the cloud-computing platform of the GEE. In this course, land cover mapping using advanced machine learning algorithms, such as Decision Trees, Random Forest, and Extra Trees in the cloud-computing platform of the Google Colab is presented. This will significantly help you to decrease the issues encountered by software and platforms, such as Anaconda. There is a much lower need for library installation in the Google Colab, resulting in faster and more reliable classification map generation. That validation of the developed models is also covered. The feature importance modeling based on tree-based algorithms of Decision Trees, Extra Trees, and Random Forest is discussed and presented. In summary, remote sensing and GIS technologies are widely used for land cover mapping. They provide accurate and timely information that is critical for monitoring and managing natural resources.Highlights:Concepts/basics of Remote SensingSatellite Image collection utilizing the Google Earth Engine (GEE)Exporting Satellite imagery into your Google DriveReference/Ground truth data creation in QGIS (vector format)Converting reference data from vector data into raster data in QGISConcepts of machine learning algorithmsReading and importing your raster and reference data from your Google Drive into Google ColabDeveloping different advanced machine learning algorithms in Google ColabLand use land cover mapping utilizing different machine learning algorithmsValidation of developed machine learning modelsFeature importance modeling using tree-based algorithmsExporting classification maps into your hard driveMap layout production in QGIS
