Forests support the biodiversity on which humanity depends, but protecting them is often at odds with the agricultural demands of the world’s growing population. Ecological features like hedgerows and linear woodland offer a potential solution for enhancing carbon storage and biodiversity without displacing crops. We developed a high-resolution, deep-learning framework to map these features across agricultural land. We used Remote Sensing Foundations’ (RSF), part of Google Earth AI, to train our AI to recognize specific features of the British countryside, and leveraged Google Earth Engine to address computational bottlenecks and scale our analysis across the whole of England. The newly released Farmscapes Vectorized dataset will empower conservationists to measure and expand these features. 

Who should attend:

• AI and Data Scientists looking for practical, high-impact applications of computer vision and foundation models.

• Ecologists and Remote Sensing Specialists interested in leveraging deep learning for environmental mapping.

• Conservationists and Policymakers seeking data-driven strategies to balance agricultural yields with biodiversity.

Why you should come:

• See a real-world blueprint for moving from an ecological problem to a deployed, national-scale AI solution.

• Learn how cutting-edge tools can bridge the gap between heavy computational demands and actionable climate tech.

• Connect with the Google scientists working at the intersection of machine learning and sustainability.

What you will learn:

• Model Training: How to leverage Remote Sensing Foundations (RSF) to train models for highly specific, localized geographic features.

• Scaling Up: Strategies for using Google Earth Engine to overcome severe computational bottlenecks when applying deep learning to massive geographic areas.

• Open Data Application: How to access and utilize the newly released Farmscapes Vectorized dataset to drive your own research or conservation efforts.