← All work
AI / Remote Sensing2026Concept - early

Sentinel: Deforestation Detection

An automated system to flag illegal deforestation from satellite imagery.

Sentinel is an early-stage concept for an automated system that combats illegal deforestation using satellite imagery. The premise is simple: forests change slowly, illegal clearing does not, and that difference is visible from orbit if something is watching for it.

The intended approach is time-series change detection over satellite tiles, surfacing suspected clearing events for review rather than trying to prosecute them automatically.

Highlights

Why change detection

Classifying a single image as forest or not is the easy part; the signal that matters is the transition from forest to cleared between two dates. Anchoring on change rather than a static label is what turns raw imagery into an actionable alert.

Stack

Data

  • Satellite imagery
  • Time series

Method

  • Computer vision
  • Change detection