Machine Learning in GIS: Land Use and Land Cover analysis

Discover how to choose stable QGIS versions by using long term releases, avoid experimental builds, and follow course materials compatible with version 3.0 across Windows, Mac, Linux, and Android.

Explore how to navigate QGIS versions and install plugins from zip files, ensuring access to stable releases and compatibility with the semi-automatic classification plugin across platforms.

Learn how to download and install QGIS on Windows, choose between long-term and latest releases, complete the setup, and launch QGIS desktop for GIS analysis.

Sign up for Google Earth Engine to access a cloud platform for remote sensing data and analysis, with a JavaScript code editor and Landsat archives for land use classification.

Install Orfeo Toolbox for image classification in GIS by following a step-by-step setup guide, then activate the toolbox in your GIS software and access the OTB panel.

Define machine learning and its link to artificial intelligence. Compare supervised, unsupervised, semi-supervised, and reinforcement learning with labeled data in GIS and remote sensing.

Learn image classification in GIS, comparing supervised and unsupervised approaches for raster remote sensing data. Assess accuracy with confusion matrices on land use maps.

Perform unsupervised classification in QGIS using a three-class k-means approach on a sentinel image to separate water, vegetation, and built-up areas.

Explore common supervised and unsupervised image classification algorithms used in GIS, including minimum distance to mean, maximum likelihood, decision trees, random forests, and support vector machines, with accuracy assessments.

Learn visual change detection as a first step to land cover analysis using the cloud-based EO Browser with Sentinel imagery, spectral indices like NDVI, and on-the-fly classifications.

learn to perform image classification with a random forest model in a GIS workflow, using training and validation data on sentinel-2 imagery to classify water, vegetation, and built-up areas.

Practice image classification using support vector machines in OTB, guided by random forest and decision trees, and prepare training and validation data to classify a separate image.

Demonstrates applying a decision tree classifier in OTB to Sentinel-2 imagery, building a training and validation workflow, producing a land cover map, and assessing accuracy with a confusion matrix.

Assess accuracy of land use and land cover maps by comparing classifications to reference data via a confusion matrix and error metrics.

Import and visualize Sentinel-2 data in Google Earth Engine for Dubai. Create a workflow to display true color and false color composites and build land cover maps with supervised methods.

Explore unsupervised image classification in Google Earth Engine for Dubai, using a means classifier, training data, and 5–10 classes, plus exporting results to drive.

Learn to perform supervised classification in Google Earth Engine using Sentinel imagery and a random forest classifier, including creating training data and merging classes.

Test accuracy of land use land cover classifications in Google Earth Engine with a Dubai example. Build a validation dataset, compute confusion metrics, and report overall accuracy.

Explore object-based image classification in QGIS with Orfeo Toolbox, covering segmentation, feature extraction, and training and validation data to generate crop maps from Sentinel composites.

Extract features for object-based image classification using zonal statistics on a Sentinel composite image, producing attribute tables from training and validation vector data to train a vector classifier.

Train vector-based classifiers for object-based image analysis using prepared training data and features, then compare random forest, decision trees, and support vector machines and assess accuracy.

Train and apply object-based crop classification in QGIS using machine learning algorithms such as random forest, SVM, and decision trees on Sentinel-derived features, validating with training data and comparing results.

Object-based image analysis combines spectral and spatial features through segmentation into objects, then classifies those objects using shape, texture, context, and statistical parameters.

Learn how to perform object-based image analysis on sentinel-2 imagery by segmenting, applying zonal statistics, extracting features, and building training, validation, and classification workflows for water, vegetation, and urban classes.

Create training data for land use classification by labeling segmented polygons as water, vegetation, or built-up, then export training and validation datasets for offline model training.

Apply object-based image classification using a random forest model on Sentinel imagery, utilizing training and validation polygons, feature fields, and accuracy assessment to produce classified land cover maps.

Add the raster classification to QGIS, apply a five-class scheme (bare soil, forest, grassland, urban, water), and customize colors and labels for each class.

Learn to convert image classification results into a final land use land cover map in QGIS using a print layout, add map, legend, north arrow, and export at 300 dpi.

Discover further opportunities in remote sensing, data science, and machine learning for GIS by exploring the instructor's Udemy courses, YouTube channel, and social pages.