Digital Event Horizon
A new study from the University of Cambridge uses satellite imagery and artificial intelligence to detect hedgehog habitats in the UK, offering a promising approach for conservation efforts.
The University of Cambridge has developed an AI model that identifies bramble patches from satellite data to map potential hedgehog habitats in the UK. European hedgehog populations have declined by 30-50% over the past decade, making it challenging to track them across large areas. The AI model combines logistic regression, k-nearest neighbors classification, and TESSERA earth representation embeddings to detect bramble patches from satellite imagery. A recent field test showed promising results, successfully locating a bramble patch within 20 seconds of receiving the model's predictions. The technique has potential implications beyond hedgehog conservation, including tracking invasive species and monitoring agricultural pests. The simplicity of the model offers practical advantages, potentially running on mobile devices for real-time field validation.
The University of Cambridge has made significant strides in using satellite imagery and artificial intelligence (AI) to map potential hedgehog habitats across the UK. Researchers have developed an AI model that identifies bramble patches, which are essential hiding spots for hedgehogs, from satellite data.
European hedgehog populations have faced a decline of approximately 30 to 50 percent over the past decade, making it challenging and expensive to track these nocturnal creatures across large areas. To overcome this challenge, researcher Gabriel Mahler developed an AI model that leverages logistic regression and k-nearest neighbors classification to detect bramble patches from satellite imagery.
The model also incorporates TESSERA earth representation embeddings, which process imagery from the European Space Agency's Sentinel satellites, with ground-truth observations from iNaturalist, a citizen science platform. This approach allows researchers to identify critical habitat features like brambles and monitor changes in ecosystems.
A recent field test conducted by Mahler and colleagues revealed promising results. During a day-long excursion, the research team successfully located their first bramble patch within 20 seconds of receiving the model's predictions. The team found substantial bramble growth in areas with high confidence scores, including Milton Country Park and Bramblefields Local Nature Reserve.
While the early experiment was an informal test rather than a formal scientific study, it demonstrates the potential for this approach to rapidly map critical habitat features. If successful, this technique could enable real-time field validation using mobile devices, allowing researchers to improve the model while verifying its predictions.
The implications of this research extend beyond hedgehog conservation. AI-based approaches combining satellite remote sensing with citizen science data could potentially be applied to track invasive species, monitor agricultural pests, and assess changes in various ecosystems. As climate change and urbanization reshape habitats for threatened species like hedgehogs, rapidly mapping critical habitat features becomes increasingly valuable.
The simplicity of the bramble detector model also offers practical advantages. Unlike more resource-intensive deep learning models, this system could potentially run on mobile devices, enabling real-time field validation. Researchers plan to develop a phone-based active learning system that would enable field researchers to improve the model while verifying its predictions.
This research application of neural network techniques serves as a reminder that AI is a diverse and complex field, encompassing not only generative models like ChatGPT but also specialized approaches like this bramble detection model. As AI continues to evolve, we can expect to see innovative solutions addressing pressing environmental challenges.
Related Information:
https://www.digitaleventhorizon.com/articles/Cambridge-Researchers-Utilize-Satellite-Imagery-to-Detect-Hedgehog-Habitats-deh.shtml
https://arstechnica.com/ai/2025/09/can-ai-detect-hedgehogs-from-space-maybe-if-you-find-brambles-first/
https://www.hashe.com/tech-news/can-ai-detect-hedgehogs-from-space-maybe-if-you-find-brambles-first/
Published: Fri Sep 26 19:01:51 2025 by llama3.2 3B Q4_K_M
