AgResearch transforms pest surveillance with AI-driven crop mapping

AgResearch in New Zealand leverages AI and remote sensing to improve biosecurity surveillance in the agriculture sector. The solution integrates high-resolution satellite imagery, drones, and Google Street View with computer vision and machine learning to map maize crops and identify biosecurity threats, such as Fall Armyworm and the invasive Tree of Heaven. Researchers combined over a million samples of satellite images with automated labeling techniques to create digital maps and inform pest management. The methodology employs LiDAR and drone photography, automatically labeling host plants using AI-driven plant identifiers like Pl@ntNet. Insights are provided to national authorities, including the Ministry for Primary Industries, enhancing surveillance strategy and resource focus. The scalable approach makes use of accessible, freely available imagery and advanced data integration to inform policy decisions. Collaboration with stakeholders, such as the Christchurch City Council Urban Forest Team, is highlighted. Impact includes informing long-term pest management, supporting sustainable agriculture and ecosystem protection, and providing a blueprint for others in biosecurity surveillance.