Machine Learning in GIS and Remote Sensing: 5 Courses in 1

Choose long-term releases of QGIS for this course to ensure stability. The video guides selecting versions, notes minor interface differences, and confirms materials work on Windows, Mac, Linux, and Android.

Learn how to install Kuji software on Windows, choosing between latest and long-term releases, and completing the install with license terms, destination path, and optional data sets.

Learn how to select and install stable or older QGIS versions to match plugin compatibility, including installing the semi-automatic classification plugin from zip via the plugin manager.

Explore the semi-automatic classification plugin for Kuji, enabling supervised classification with image preprocessing, training input creation, regions of interest, change detection, and accuracy assessment, backed by documentation and case studies.

Learn how to install and manage QGIS plugins, including version considerations, and explore examples like the Sentinel plugin and the semi-automatic classification plugin for remote sensing analysis.

Explore machine learning fundamentals for gis and remote sensing, including data-driven learning from labeled and unlabeled data, and supervised, unsupervised, semi-supervised, and reinforcement learning with classification and regression examples.

Explores how artificial intelligence, machine learning, and deep learning apply to GIS and remote sensing, including classification, clustering, prediction, and computer vision with satellite and drone imagery.

Explore image classification in GIS and remote sensing using supervised and unsupervised approaches. Build training sets, apply algorithms like random forest and SVM, and evaluate accuracy with confusion matrices.

Download Landsat imagery via the semiautomatic classification plugin in QGIS, using USGS or NASA or Copernicus portals, set study coordinates, pick dates and cloud limits, preview bands.

Explore supervised and unsupervised image classification algorithms, including Kamins clustering, minimum distance to mean, maximum likelihood, Bayesian, decision trees, random forests, and support vector machines, with examples and accuracy considerations.

Learn to perform visual change detection and land cover classification in the EO browser using Sentinel imagery and spectral indices, with qualitative inspection as a first step before quantitative analysis.

Explore regression analysis to estimate relationships between dependent and independent variables, using linear and multivariate models for prediction and causal inference, with examples like crime hotspots and disease prevalence.

Explore geospatial predictive modeling with machine learning and deep learning to forecast events in GIS using environmental and socio-cultural factors; apply regression and interpolation techniques to big data.

Explore image classification in ArcGIS, covering unsupervised and supervised pixel-based methods, as well as object-based approaches, with hands-on steps for training samples, signature files, and post-classification cleanup.

Explore ArcGIS Pro desktop in this course, with emphasis on free trial access, version 10.6 or higher compatibility, and tips to locate functions via search plus available help resources.

Perform unsupervised image classification in ArcGIS using a multi-band Sentinel composite. Set class count and sampling to reveal agricultural fields, harvested fields, and built-up areas for subsequent supervised classification.

Follow a step-by-step guide to install the orfeo toolbox plug-in for qgis, activate the toolbox panel, and access machine learning algorithms for image classification.

Apply unsupervised classification in QGIS using k-means clustering to segment a city image into three classes—water, vegetation, and built-up—adjusting input, output, and rendering results.

Define mutually exclusive, exhaustive land cover classes and build representative training data for supervised LULC classification. Perform pixel-wise classification to produce maps, then assess accuracy to validate results.

Select training data in ArcMap 10.6 by outlining cropland, water, trees, and settlements with polygons, label each area, and save the training samples to build a supervised classification model.

Apply supervised image classification for land use land cover with a support vector machine in ArcGIS, using training data and local processing; compare SVM with random forest results.

Perform land use and land cover classification on a Landsat 5 image using the maximum likelihood algorithm, with training data and polygons, and assess results with color coded classes.

Develop a supervised lulc map from Landsat 5 imagery in qgis by applying atmospheric correction and a semi-automatic classification plugin. Build training data for vegetation, soil, and water.

Run a land use land cover classification with the minimum distance algorithm on forests, soil, and water, compare with maximum likelihood, and note that visual assessment favors maximum likelihood.

Explore accuracy assessment of image classification in GIS using confusion matrices and reference data. Learn visual and quantitative controls, and compute overall accuracy, user accuracy, and producer accuracy.

Create validation data and perform accuracy assessment for land use land cover maps from Landsat imagery, applying visual checks and numeric validation across vegetation, soil, and water classes.

Perform LULC accuracy assessment in QGIS with the semi-automatic classification plugin, using training and validation data to generate an accuracy report showing producer, user, and overall accuracies.

Learn to perform random forest supervised classification of Sentinel-2 imagery in QGIS by building a model with training and validation data and generating a classified land-cover map.

Compare decision tree and random forest classifiers for land cover from Sentinel-2 imagery, detailing training, validation, model accuracy metrics, and visual difference in results within a QGIS/OTB workflow.

Learn how images are segmented for GIS and remote sensing, showing how adjacent pixels form segments via spectral details to improve classification.

Discover how to download image data for segmentation analysis in GIS and remote sensing using the semi-automatic classification plugin from USGS, NASA, and Copernicus portals.

Learn to perform image segmentation in ArcGIS with the segment mean shift function, adjust spectral detail and spatial detail, and export a segmented layer for object-based image analysis.

Learn to perform image segmentation in kuji (qgis) using the otb toolbox on sentinel imagery, applying the mindshift algorithm, and exporting a segmentation as a shapefile for object-based image analysis.

Explore object-based image analysis (OBIA) that segments imagery into objects and classifies them using spectral, shape, texture, and context, contrasting OBIA with pixel-based methods for multi-scale landscapes.

Shows how to create training data for object-based image classification in ArcGIS by selecting segments from a segmented raster and collect 15–20 samples per class, such as white roofs.

Learn to apply object-based image classification (OBIA) in ArcGIS using training data and a signature file. Build and apply a support vector machine classifier to produce a final classified image.

Master the regression analysis equation, identifying the dependent variable, explanatory variables, coefficients, and residuals, and follow a GIS workflow from problem framing to six statistical checks in exploratory regression.

Explore ordinary least squares regression in GIS, perform six validation checks: coefficient significance, non-stationarity, sign consistency, multicollinearity via VIF, residual bias and normality, autocorrelation; compare models using R-squared and AIC.

Apply regression analysis to geospatial data, build a properly specified model, assess six checks including coefficients, variable inflation factor, residual autocorrelation with Moran's I, and interpret biomass drivers.