AI-Powered Satellite Imagery Maps Cashew Farms Across Guinea-Bissau to Combat Deforestation

Researchers have developed an innovative remote sensing approach utilizing machine learning and Sentinel-2 satellite imagery to accurately map cashew orchards across Guinea-Bissau. This initiative addresses a critical need for nationwide data on cashew production, which is a major economic activity in the country and wider West Africa but contributes to deforestation and biodiversity loss when unregulated.

The study employed margin-based Active Learning techniques to optimize the training of the machine learning model, achieving a balanced accuracy of 94.0% in detecting cashew orchards. This entirely off-site methodology offers a scalable and cost-effective solution for monitoring agricultural land use.

The project has made two datasets and a high-resolution 2021 cashew map openly accessible via GitHub. This open-source contribution provides a vital resource for environmental monitoring and sustainable agricultural planning, offering a foundational tool for broader environmental applications in the region.

For Guinea-Bissau and other West African nations, this technology represents a significant step towards managing natural resources more effectively. By providing precise data on cashew cultivation, it can help policymakers and environmental organizations mitigate the negative impacts of unregulated farming practices, promote sustainable agriculture, and protect valuable ecosystems.