Machine-learning cameras developed by Canberra-based Trellis Data have been used at Australian ports to help detect invasive species hiding in cargo containers, without having to manually inspect them.

The technology has been used as part of a two-week trial by the Department of Agriculture at the Port of Brisbane.

The cameras, which scan the container from multiple angles in just four seconds, analysed more than 1.7 million images of containers during the fortnight, with 4 per cent of containers having pests that needed to be removed.

Trellis Data's product manager James Meszes said the technology had the potential to be expanded to other port or freight facilities across the country.

"It's about taking something that is near impossible as a human, to be able to detect small objects while containers are being constantly being moved, while also dealing with rain and sun glare and rust on the containers," Mr Meszes said.

"This technology is there to assist and provide assurance that every container can be scanned at every crane at every port, and it's there to work side-by-side with people.

"The Department of Agriculture is only able to inspect about 3 per cent of all goods manually, so the cameras can help to solve this problem."

Known as the Biosecurity Automated Threat Detection System, the cameras scan the outside of the cargo container to detect any pests that may have 'hitchhiked' from overseas.

The cameras are able to detect invasive species smaller than 10 millimetres in length.

Last year, 2.5 million shipping containers came into Australia from overseas.

READ MORE:

If the rate of pest detection picked up by the cameras remained consistent, that would mean 100,000 containers would have required treatment to get rid of the invasive species hiding onboard.

Mr Meszes said the technology also used machine learning to help identify the species of foreign pests that may be hiding among cargo goods.

"We've structured it so it can be trained and improved upon, so if there was a pest that was an issue overseas, and there are some photos taken of it, those images can be used to identify it at every port that Australia has, so that capability is also in place," he said.

"At the end of the day, we want to be able to protect Australia and its agriculture industry."

It's estimated an outbreak of invasive species such as khapra beetles or giant African snails could cost billions of dollars in damages to the Australian agricultural sector.

The new camera technology comes just months after a khapra beetle outbreak hit parts of Canberra.

The invasive beetle species was discovered in a shipment of fridges, which forced the closure of several Good Guys stores in the capital, as biosecurity officers sought to prevent the pest from spreading.

Mr Meszes said the camera technology could help to save 100,000 hours of effort every year for the Department of Agriculture.

"This is something that machine learning is perfect for, and we can only improve as it goes down the line," he said.

"These are pests that hard to spot, but this can do it in real time."

Work on the project had been ongoing for almost three years before the trial started, but Mr Meszes said the technology could only get better with more containers being detected.

A report on the two-week trial is currently being finalised, and will help to determine whether the technology would be able to be used on an ongoing basis in Brisbane, or in more locations across the country.

While the technology is being used at ports, Mr Meszes said it could also be transferred to other settings such as airports, or other areas where goods come into the country.

"We do a lot of work with object detection, and anywhere there's a camera, we can use it," he said.

A spokesman for the Department of Agriculture said the results of the two-week trial had been impressive, and that further tests were expected to be undertaken.

"The technology being considered could reduce inspection times by identifying and informing the inspector where a threat, such as a live pest, is located on a container," the spokesman said.

"There is potential to adapt the technology for inspection of areas that are difficult to access, for example when conducting tailgate inspections or large consignments of palleted goods."

Our journalists work hard to provide local, up-to-date news to the community. This is how you can continue to access our trusted content: