Remote Sensing Imagery and ArtificiaI Intelligence
Robert Morrison
Principal Consultant at Esri Ireland
Remote Sensing Imagery and Artificial Intelligence
Radio Frequency Tracking
3D Buildings
Telco
Modeling
Vegetation Visualization
UAV Data
Capture
Lidar-Based Topo Map
Elevation and
Feature Extraction
Wildfire Imagery
Feature Extraction
Hydrology Analysis
Hurricane Damage
Florida
Wildfire
Aerial Imagery
Neural Networks
TensorFlow
CNTK
Natural Language
Processing
Cognitive
Computing
GeoAI
Computer Vision
Object
Detection
Support Vector Machines
Object Tracking
Theano
scikit-learn
T-SNE
Random Forest
Caffe
Machine
Learning
Deep
Learning
Artificial Intelligence
Computer does a task with
some level of human
intelligence
Techniques that learn from
data
Loosely resembles the
human brain
Machine
Learning
Deep
Learning
Artificial Intelligence
Image Classification Object Detection Semantic Segmentation Instance Segmentation
Image Use Cases
GIS Includes Machine LearningTools
GIS
Classification
Clustering
Prediction
Deep Learning
Precision agriculture Asset management
Suitability analysis Change detection
Imagery contains ‘data’ - Machine learning extracts ‘information’
Fleet management
Agricultural yield
Target identification
Machine Learning and Imagery
Machine
Learning
GIS
Derived structured
information products
Imagery
Access
Imagery
Prep
Creating
Training
data
Inference
Distributed
Processing
Feedback
Loop
Take
Action
AI with GIS: End to End Cycle
Training Derive
Products
Case Study
Palm Tree Health
- Tonga is a collection of 170 islands in the South Pacific
- Economy is heavily dependent on coconuts
- Efficient management of coconut plantations is vital
What if, instead of manually visiting every tree and
inspecting, we used imagery to automate the process?
Coconut Palm Health in the Kingdom of Tonga
Kolovai Palms
Train the data
Export the training data
Execute the deep learning model
Execute the deep learning model
Assess individual palm health
Derived Information Product
1. Create a point for each palm.
2. Buffer each point by 3m.
3. Intersect with TGI imagery.
4. Calculate zonal statistics.
Deep Learning Examples
Impervious Mapping Change Detection
Flood Planning
Riparian CorridorsLandcover Classification
and more…
Planning Compliance
Environment Compliance
Emergency Management
Robert Morrison
rmorrison@esri-Ireland.ie

Remote Sensing Imagery & Artificial Intelligence