Project SPARROW Takes Off
Written by Lucy Black   
Friday, 28 November 2025

Fundación Biodiversa in Colombia has become the first pilot of Microsoft's Project SPARROW. SPARROW, developed by Microsoft's AI for Good Lab, is an AI-powered edge computing solution designed to monitor and protect wildlife in the most remote regions of the world.

The key features of SPARROW start with the power management. It uses solar charging, battery health monitoring, and dynamic component scheduling with the aim of achieving continuous off-grid operation. The power is used for camera traps, acoustic monitoring, and environmental sensors. The system has on-device AI that runs optimized PyTorch Wildlife models on power efficient edge GPUs such as the Jetson Orin Nano for real-time image and acoustic detection, species classification, and event recognition.

SPARROW uses AI to identify species through automatic classification, and for animals that have suitably individual appearances, like the patterns on zebras or giraffes, the AI can re-identify the same animals, helping researchers measure survival rates and populations.

The new pilot is taking place with the Fundacion Biodiversa team at the El Silencio Natural Reserve and Research Station, located within MMV in Colombia. The foundation is a nonprofit organization dedicated to preserving Colombia's biological and cultural heritage. It is using SPARROW to collect biodiversity data, which is processed locally using Microsoft's PyTorch-based wildlife AI models running on low-energy edge GPUs. SPARROW maintains a link to the cloud or on-premise infrastructure through low-Earth-orbit satellites meaning the data is available in near real-time. When the units can't be online, SPARROW records the data and automatically synchronizes with the central store once connectivity is restored to ensure no loss of information.

Until now, the local rangers carried out passive monitoring of camera and acoustic traps, but faced difficulties in analyzing the massive amounts of data the cameras and microphones gather. In addition, since many of the cameras and microphones are in remote locations, checking them frequently to make sure they're working properly isn't always possible. 

The use of SPARROW means that, instead of spending hours analyzing and organizing, researchers have passed that task over to AI, giving the rfesearchers more time to understand what the data means. 

The SPARROW client is available on GitHub. 

 sparrow

More Information

SPARROW Client on GitHub

SPARROW Ted Talk

Related Articles

Microsoft Launches Human Centered AI Tool

Reconstructing Banknotes Using Computer Vision

 

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Banner


Deepnote Goes OpenSource
18/11/2025

Deepnote, hailed as the Jupyter Notebook successor,
is now open-sourced after 7 years running as a sole cloud offering.



Kotlin 2.3 Improves Swift Interop
27/11/2025

Kotlin 2.3 is available now as a release candidate. The new version adds a new checker for unused return values, and changes to context-sensitive resolution. The release candidate adds support for Jav [ ... ]


More News

pico book

 

Comments




or email your comment to: comments@i-programmer.info