Thursday, November 30, 2023

AI in the sky closing in on pest plants

Avatar photo
Successful ECan pilot programme shows promise against Nassella tussock.
Pest plant coverage as predicted by AI aerial imagery in this Nassella tussock trial may point to the way of the future for landowners planning pest-control programmes.
Reading Time: 3 minutes

Environment Canterbury (ECan) has taken its first steps towards using artificial intelligence to identify pest plants from aerial imagery.

A successful pilot programme in South Canterbury has shown that AI may provide a more efficient way to inspect large swathes of land for biosecurity threats such as Nassella tussock.

The tussock is an invasive grass that is rapidly spreading in Canterbury. It displaces productive pastures and is unpalatable to stock. 

Only 70 years ago, Nassella drove some farmers off the land after it ran rampant, taking over entire farms, leaving no room for crops or stock.

Each year, ECan’s biosecurity team spend months inspecting properties and ensuring landowners identify and control this pest plant and control its spread – a requirement of the Canterbury Regional Pest Management Plan.

New technologies now have the potential to streamline pest plant control while increasing productivity both for the regional council’s work programmes and for affected landowners.     

ECan recently partnered with Wellington-based Lynker Analytics to develop and train an AI programme to identify the pest plant from aerial imagery. 

Lynker Analytics has specialist expertise in data science, data infrastructure and geospatial analytics.

High-resolution imagery, collected by a small plane, was captured with the support of the landowner.

Lynker Analytics managing director Matt Lythe said the test image relies on high-quality images. 

The photo used in the trial was 13GB in size and contained 3.6 billion pixels.

The machine learning model was trained by splitting up the large image into non-overlapping 512×512 pieces, called train, validation and holdout sets.

“This allowed the programme to learn over time to identify a pixel or group of pixels as either Nassella or not-Nassella,” Lythe said.

The benefits for the regional council in having AI identify pests through imagery are greater efficiencies in identifying previously unknown infestations, and in monitoring. It could potentially be used for initial inspections too, reducing the need for biosecurity staff to physically inspect the land as often.

There is still a while to go before it can be implemented across the region, but landowners may in future be able to use the AI to ascertain infestation levels on their own properties and plan control programmes accordingly, Lythe said.

ECan biosecurity team leader Matt Smith said the initial results show that the AI has a 90% success rate in  identifying large, mature Nassella tussock. 

More refining of the model is required to improve that identification rate.

“An AgResearch study found that hand-control operations remove about 30-35% of Nassella tussock, mostly very small plants, which means about 65-70% is left behind,” Smith said.

A mature Nassella tussock can produce up to 100,000 seeds, so it is important to remove all infestations prior to viable seed being produced each year.

Currently ECan’s Nassella tussock programme is mostly carried out on foot.

Smith said there could be many benefits via AI to wider biosecurity programmes in future, including monitoring pest trends over time.

“In biosecurity, things have been done the same way for a long time and now we have this amazing technology that is becoming available and we are looking to the future. 

“It’s a big part of how we can protect the region going into the future.

“We’re embracing new technologies for biosecurity, especially when they can increase productivity both for our work programmes and for affected landowners,” Smith said.

Lythe said geospatial data, technology and analytics are core competencies in the business that provides a wide range of farm paddock and environmental  services including monitoring riparian plant survival, measuring carbon sequestration and monitoring deforestation and replanting.

Geospatial analytics is the discovery, interpretation and communication of patterns within geospatial data in order to promote better business decision making. 

“A geospatial system of record is essential to good organisation outcomes along with a robust data management plan and data architecture to help solve complex problems through the power of geography.

“Data visualisation is really the modern equivalent of visual communication, building engagement systems through data visualisation to explain the significance of information by placing it in a visual context.

“It can be used to equally monitor farmland for good plantings as well as the unwanted pest plants.”

Lynker Analytics is currently working with Silver Fern Farms undertaking sequestration mapping for its net zero carbon project.

“Nassella is a only model and it is quite effective to train to be a very dedicated model for other work such as this net zero carbon project,” Lythe said.

People are also reading