← back
Environmental monitoring

Project Jacaranda

No company is born ready. Some are born from a satellite pointed at the Amazon.

2022

Finalists in Tech4Positive Futures, a global ESG award, with a project that used convolutional neural networks and public satellite imagery to detect deforestation in the Amazon. Affordable, efficient, and open source.

Since then, we adapted the model. And it proved applicable to far more than a forest. The results were promising enough to make it unacceptable for the story to stop there.

BRISE is what came next.

Video case
Tech stack
Convolutional neural networks
Model trained for land cover classification from multispectral satellite imagery.
Public satellite imagery
Sentinel-2 and Landsat as primary sources. Sufficient resolution for regional-scale deforestation detection.
Open source
Reproducible pipeline: from image acquisition to classification output. No dependency on proprietary APIs.
Low operational cost
Free data + lightweight inference. Designed to be viable in budget-constrained contexts.

This project gave birth to BRISE.

The model proved applicable far beyond forests. See what came next →