Boosting biosecurity surveillance: The marriage of remote sensing and AI

Research, led by senior scientist Dr Federico Tomasetto, uses remote sensing technology (in this example satellites) to identify maize crops.  

The satellites images of rural New Zealand taken from space are then combined with semi-automatic, image labelling techniques using computer vision. They are then integrated with more than 1 million samples of multi-temporal high-resolution satellite images associated to 16 crop types.  

Once this process, which is power by AI is complete, Federico has been able to glean invaluable insights into maize crop locations, and by combining this pixel-based dataset, Federico and his team can provide stakeholders with: 

• Digital maps of where New Zealand maize crops are grown 
• Information on maize trends: where new maize crops are being planted  
• Data on locations most at threat to incursions  
• Advice to authorities so they can swiftly respond to emerging threats  

This Better Border Biosecurity (B3) funded research has been shared with the Ministry for Primary Industries and the Foundation for Arable Research so they can fine-tune their surveillance efforts.  

Federico explains: “The data we have been able to generate will help with decisions on where we need to focus our efforts to control pests such as Fall Armyworm. It arrived in New Zealand early in 2022 and is now widespread particularly in the North Island. One of the main host plants for Fall Armyworm is maize and sweetcorn so our research is timely and will provide everyone with much more precise and accurate data on where the pests main host plants are and help focus on long-term management of the pest by industry. We could not have done it without AI, and we are now looking at other ways we can use large datasets to tackle other biosecurity threats here in New Zealand.”

Host plants are a common way of identifying and zeroing in on biosecurity threats. Fall Armyworm congregate in maize crops. As industries like agriculture navigate an increasingly interconnected global landscape, the need for innovative solutions to safeguard ecosystems, such as AI, are also being used to uses clues and identify other pests too.  

Federico and his team are also combining Google Street View and photos taken from drones with state-of-the-art AI algorithms to remotely identify specimens of the invasive Tree of Heaven (TOH), a notorious host of pests like the brown marmorated stink bug. [and spotted lanternfly]

A hoc workflow has been deployed to search Christchurch city streets for TOH integrating Light Detection and Ranging (LiDAR) technology, drone images that are fed into the plant identifier called, Pl@ntNet, that automatically labels the tree.  

This method of using AI and Google Street View (GSV) is becoming increasingly popular in cities worldwide. The Christchurch City Council Urban Forest Team is interested in the research and Federico and his team have been able to provide them with insights into where the tree is becoming more prevalent in areas around the city.

What sets this approach apart is its accessibility and scalability, relying on freely available imagery and data sources to inform decision-making. By harnessing the collective power of remote sensing and AI, Tomasetto's work represents a change in thinking in biosecurity surveillance, offering a blueprint for sustainable pest management practices.