Agricultural Monitoring and Food Security

Crop mapping with foundation models, vegetation stress detection, and honest ground-truth validation

Agricultural Monitoring and Food Security
Intermediate Geospatial AI 2 lessons Samuel Appiah Kubi

About this course

Applying satellite data to agriculture at the scale a food-security programme needs — across Ghana's cocoa belt, Sahel smallholder farms, and Nile Delta irrigated land — using a modern foundation model for crop classification, classic vegetation indices for stress detection, and a validation methodology that reports accuracy honestly rather than cherry-picking a favourable number.

What you'll learn

  • Run Prithvi-EO-2.0's crop mapping head across 10 USDA-calibrated crop classes
  • Detect vegetation stress early using red-edge indices, before it's visible in plain NDVI
  • Monitor irrigated area expansion using NDWI time series
  • Validate a model's output against real field-collected ground truth
  • Understand why overall accuracy alone is a misleading metric on imbalanced crop classes

Requirements

  • Optical Analysis with Sentinel-2
  • An introduction to pygeovision's AI models

Course content

Prithvi-EO-2.0 for Crop Classification Preview 30 min
Validating Against Ground Truth, Honestly 24 min
Free
Login to Enroll
  • LevelIntermediate
  • Lessons2
  • CertificateYes
  • AccessLifetime
Samuel Appiah Kubi Instructor