Researchers at the University of Cambridge have developed an innovative artificial intelligence model that allows for the identification of potential hedgehog habitats in the UK using satellite imagery. Since these animals are too small to be directly seen from space, specialists focused on analyzing bramble thickets—where hedgehogs find shelter and food sources.
This is reported by Finway
Satellite Technology and Machine Learning for Biodiversity Protection
To implement the project, researchers used data from the European Space Agency’s Sentinel satellites in combination with machine learning algorithms. The model integrates logistic regression methods, nearest neighbor analysis, and the TESSERA system for processing satellite images. This complex was supplemented with observations from citizen scientists collected through the iNaturalist platform.
This combined approach allowed for the creation of a detailed map of probable hedgehog habitats across the UK. To verify the model’s accuracy, field trials were conducted in Cambridge, where researchers compared the AI’s predictions with real data, confidently identifying large open bramble thickets. However, detecting smaller bushes under trees remained more challenging due to the limitations of satellite imagery.
Opportunities for Ecosystem Monitoring and Future Development
Although the project is still in its early stages, the authors consider it a promising alternative tool for monitoring populations of vulnerable species. Satellite analytics can cover large areas, making it significantly more efficient than traditional labor-intensive nighttime observations. The scientists emphasized that this is currently a proof of concept and the model has not yet undergone full scientific peer review. However, the team plans to expand testing and develop an active learning system that can be used in the field via mobile devices.
“University representatives highlighted that the potential applications of the method extend far beyond hedgehog protection. Similar algorithms could be used for monitoring invasive plants, agricultural pests, or tracking changes in ecosystems.”
The project demonstrates how modern artificial intelligence tools can effectively complement traditional field research methods and contribute to biodiversity conservation.