Advanced Machine Learning and Deep Learning for Geospatial Data Analysis

Module 1: Geospatial Data Foundations and Python for GIS

ü  Overview of Remote Sensing and GIS Data Models (Raster vs. Vector)

ü  Python Environment Setup for Spatial Analysis (Anaconda, GDAL, Fiona)

ü  Handling Vector Data with GeoPandas

ü  Working with Raster Data using Rasterio and NumPy

ü  Practical Session: Reading, visualizing, and manipulating satellite image bands using Python libraries

Module 2: Advanced Remote Sensing Data Pre-processing

ü  Radiometric and Atmospheric Corrections

ü  Orthorectification and Geometric Correction

ü  Image Fusion, Stacking, and Mosaicking

ü  Multi-spectral Indices (NDVI, NDWI) Calculation

ü  Practical Session: Correcting satellite imagery and calculating specialized vegetation indices

Module 3: Feature Engineering for Geospatial Data

ü  Creating Spatial Features (Texture, Edges, Shape Indices)

ü  Tasseled Cap Transformation and Principal Component Analysis (PCA)

ü  Handling NoData Values and Outliers

ü  Zonal Statistics and Feature Aggregation

ü  Practical Session: Using PCA to reduce the dimensionality of multi-spectral imagery

Module 4: Fundamentals of Traditional Machine Learning

ü  The Machine Learning Workflow: Training, Validation, Testing

ü  Supervised vs. Unsupervised Learning

ü  Introduction to Scikit-learn for Machine Learning

ü  Feature Scaling and Data Splitting for Spatial Data

ü  Practical Session: Implementing K-Nearest Neighbors (KNN) for a simple classification task

Module 5: Supervised Classification: Model Selection and Training

ü  Decision Trees and Random Forest Classifiers

ü  Support Vector Machines (SVM) for Remote Sensing

ü  Hyperparameter Tuning and Cross-Validation

ü  Training a Model using Spectral and Engineered Features

ü  Practical Session: Training a Random Forest model for land cover classification on a Sentinel-2 image

Module 6: Unsupervised Clustering and Segmentation

ü  K-Means and DBSCAN Clustering for Image Segmentation

ü  ISODATA Clustering for Initial Land Cover Mapping

ü  Image Segmentation Techniques (e.g., watershed)

ü  Labelling Unsupervised Clusters for thematic maps

ü  Practical Session: Performing K-Means clustering on an image and interpreting the resulting segments

Module 7: Introduction to Deep Learning and Neural Networks

ü  Fundamentals of Artificial Neural Networks (ANNs)

ü  Activation Functions, Layers, and Backpropagation

ü  Introduction to TensorFlow/Keras for Deep Learning

ü  Data Preparation for Deep Learning Models

ü  Practical Session: Building and training a simple dense neural network for classifying points of interest

Module 8: Convolutional Neural Networks (CNNs) for Imagery

ü  The Role of Convolutional Layers, Pooling, and Filters

ü  Designing CNN Architectures for Image Analysis

ü  Transfer Learning using Pre-trained Models

ü  Handling Large Remote Sensing Image Patches

ü  Practical Session: Building a basic CNN to classify small image chips (e.g., different crop types)

Module 9: Semantic Segmentation for Land Cover Mapping

ü  Understanding Semantic Segmentation vs. Classification

ü  U-Net and Fully Convolutional Networks (FCNs) Architecture

ü  Pixel-level Classification and Labelling

ü  Creating High-Accuracy Land Use/Land Cover (LULC) Maps

ü  Practical Session: Implementing and training a U-Net model for high-resolution building footprint extraction

Module 10: Object Detection in High-Resolution Imagery

ü  Difference between Classification, Segmentation, and Object Detection

ü  Introduction to YOLO (You Only Look Once) and R-CNN architectures

ü  Creating Bounding Boxes and Annotation Data

ü  Applications in Vehicle Counting and Infrastructure Monitoring

ü  Practical Session: Using a pre-trained model for object detection (e.g., ships in a harbor) and visualizing results

Module 11: Time-Series Analysis and Change Detection

ü  Analyzing Multi-temporal Remote Sensing Data Stacks

ü  Techniques for Change Detection (Image Differencing, Post-Classification Comparison)

ü  Recurrent Neural Networks (RNNs) and LSTMs for Sequential Data

ü  Machine Learning for Forecasting Environmental Variables

ü  Practical Session: Detecting deforestation over a five-year period using time-series analysis

Module 12: Model Evaluation and Spatial Accuracy Assessment

ü  Confusion Matrices and Classification Accuracy Metrics

ü  Kappa Coefficient and F1 Score for Model Performance

ü  Spatial Accuracy Metrics (Producer's, User's Accuracy)

ü  Techniques for Spatial Cross-Validation

ü  Practical Session: Calculating and interpreting the accuracy report for a LULC map

Module 13: Advanced Vector Data Machine Learning

ü  Machine Learning on Attributes of Vector Features

ü  Spatial Regression Models and Geographically Weighted Regression (GWR)

ü  Predicting Spatial Phenomena (e.g., pollution levels)

ü  Feature Selection for Vector-based Models

ü  Practical Session: Using Machine Learning to predict property values based on proximity features

Module 14: Integrating Lidar Data with Deep Learning

ü  Handling Point Cloud Data Structures

ü  Feature Extraction from Lidar (Height, Intensity)

ü  Voxel-based and PointNet Architectures (Conceptual)

ü  Classification of Lidar Point Clouds (e.g., ground vs. non-ground)

ü  Practical Session: Preparing Lidar data features for a classification model

Module 15: Cloud Computing for Machine Learning Remote Sensing

ü  Utilizing Google Earth Engine (GEE) for scalable analysis

ü  Running Deep Learning models on cloud GPU environments (AWS/Azure)

ü  Handling Petabytes of Geospatial Data Efficiently

ü  Serverless Computing Concepts for Remote Sensing Workflows

ü  Practical Session: Running a Machine Learning classification algorithm on a large area using GEE

Module 16: Operationalizing and Deploying Geospatial Models

ü  Saving and Loading Trained Models

ü  Integrating Models into Web GIS Applications (APIs)

ü  Creating Automated Prediction Pipelines

ü  Monitoring and Maintenance of Geospatial AI Systems

ü  Practical Session: Deploying a simple classification model via a lightweight web service (Flask/FastAPI concept)

Module 17: Geospatial Ethics and Bias in Satellite Imagery

ü  Data Privacy and Security Concerns with High-Resolution Imagery

ü  Bias in Training Data and its impact on Model Fairness

ü  Responsible Use of Geospatial AI

ü  Legal and Regulatory Frameworks for Remote Sensing Data

ü  Practical Session: Analyzing a training dataset for potential spatial or demographic bias

Module 18: Capstone Project: End-to-End Geospatial AI Solution

ü  Independent Project Planning and Data Acquisition

ü  ü  Participants should be reasonably proficient in English. 

ü  Applicants must live up to Armstrong Global Institute admission criteria.

Terms and Conditions

ü  Discounts: Organizations sponsoring Four Participants will have the 5th attend Free

ü  What is catered for by the Course Fees: Fees cater for all requirements for the training – Learning materials, Lunches, Teas, Snacks and Certification. All participants will additionally cater for their travel and accommodation expenses, visa application, insurance, and other personal expenses.

ü  Certificate Awarded: Participants are awarded Certificates of Participation at the end of the training.

ü  Course Improvement: The program content shown here is for guidance purposes only. Our continuous course improvement process may lead to changes in topics and course structure.

ü  Approval of Course: Our Programs are NITA Approved. Participating organizations can therefore claim reimbursement on fees paid in accordance with NITA Rules.

Booking for Training

Simply send an email to the Training Officer on training@armstrongglobalinstitute.com  and we will send you a registration form. We advise you to book early to avoid missing a seat to this training. Or call us on +254737296202 / +254725012095 / +254724452588

Payment Options

We provide 3 payment options, choose one for your convenience, and kindly make payments at least 5 days before the Training start date to reserve your seat:

ü  Groups of 5 People and Above – Cheque Payments to: Armstrong Global Training & Development Center Limited should be paid in advance, 5 days to the training.

ü  Invoice: We can send a bill directly to you or your company.

ü  Deposit directly into Bank Account (Account details provided upon request)

Cancellation Policy

ü  Payment for all courses includes a registration fee, which is non-refundable, and equals 15% of the total sum of the course fee.

ü  Participants may cancel attendance 14 days or more prior to the training commencement date.

ü  No refunds will be made 14 days or less before the training commencement date. However, participants who are unable to attend may opt to attend a similar training course at a later date or send a substitute participant provided the participation criteria have been met.

Accommodation and Airport Transfer

For physical training attendees, we can assist with recommendations for accommodation near the training venue. Airport pick-up services can also be arranged upon request to ensure a smooth arrival. Please inform us of your travel details in advance if you require these services. For reservations contact the Training Officer on Email: training@armstrongglobalinstitute.com or on Tel: +254737296202 / +254725012095 / +254724452588