Skip to content
menu icon

GRDC Websites

AI delivers step-change in grain quality

Agriculture Victoria senior researcher Dr Joe Panozzo is developing machine imaging technology for use in the grains industry.
Photo: Andrew Cooke

A GRDC co-investment in the application of machine vision technology for assessing pulse quality traits is bearing fruit at the Horsham Grains Innovation Park, offering high-speed analysis and the prospect of objective testing and other industry uses.

The joint GRDC-Agriculture Victoria project, 'Pulse Bio 2: Quantifying the value of pulses', utilises high-throughput digital imaging equipment to capture thousands of images of grain travelling along a conveyor.

The project has developed a series of algorithms based on the images to define grain characteristics such as size, shape, volume, colour and defects in pulses, including lentils, chickpeas, field peas and faba beans.

"The objective is about developing high-throughput measurement systems that quantify the value of pulses through traits important for the export market," says Dr Joe Panozzo, the senior researcher who leads Agriculture Victoria's seed phenomics and quality traits laboratory.

Dr Panozzo says the algorithms developed by the project have potential for use in the grain marketing chain.

"Within the pulse industry, the quality of pulses is simply defined and traded on visual characteristics such as seed size, shape, colour and absence of defects," he says.

"However, these assessments, particularly for defects and colour, are often subjective.

"The adoption of machine vision offers an objective alternative that is applicable either at the point of grain-receival or in an international market.

"While the application is currently for plant breeding, hopefully the grains industry will adopt the technology as an objective measurement of quality when a grower delivers grain to a receival point, rather than rely on colour-reference charts."

Trait analysis

From a plant breeding perspective, the technology has made grain trait analysis much faster and cheaper.

"Machine imaging means we don't need to do the laborious assessments, such as sieving of grain, to determine seed size distribution or equally time-consuming methods for measuring colour and defects," Dr Panozzo says.

About 12,000 germplasm lines are tested for quality traits at the Agriculture Victoria laboratory in Horsham each year.

"We have now applied high-throughput, non-destructive testing which we can apply to selecting germplasm with optimal quality traits," Dr Panozzo says.

"Previously, we weren't able to test the grain produced by all of our trials in the window of time between harvest and sowing the following season.

"The only data that we really had was yield data and agronomic data; however we could never complete all of the quality data in time."

This meant that breeders often did not have access to enough information when choosing germplasm.

"We were sowing lines which may not have the optimal qualities, so there was a level of inefficiency," Dr Panozzo says.

Dr Panozzo estimates that objective measurement technology has cut a year off the breeding process.

More information: Dr Joe Panozzo, joe.panozzo@ecodev.vic.gov.au

back to top