Biosecurity Innovation Program

BMSB app interview

Dr Michael Elias, Australian National Insect Collection:

So stink bugs are quite a diverse group in Australia. There are about 600 different native species in Australia and potentially thousands of species we haven't discovered yet.

Generally speaking, they're quite specialist, so they'll only eat one or two species of plants. And because they've evolved here, they'll only eat native species.

There are a few which are more generalist, and they tend to be the bigger pests. But the biggest pests are the ones which come from overseas because they adapted to eat crops. Most of our crops come from somewhere else, so they are adapted to eat those crops, and they have no natural predators.

Brown Marmorated Stink Bug is originally from East Asia, especially China, especially the weather regions of China. Because it's such a generalist species, it targets a very wide variety of crops, especially stone fruit, apples, pears, tree nuts--especially hazelnuts, it's a big, big pest on hazelnuts--and wine grapes, and there's several other smaller crops as well.

The way they deal with most pests when they find them at the border is they go out and collect them. Because they're not trained entomologists, they send them back to an entomology lab. In the meantime, that cargo has to be held in quarantine at the expense of the importer. If the biosecurity officers had a tool, which they could use to identify them on the spot, you wouldn't need to spend that much time, which is a lot of money for various importers.

Dr Alexander Schmidt-Lebuhn, CSIRO Scientist, Centre of Australian National Biodiversity Research:

This is brown moderated stink bug, which is a major biosecurity threat for Australia. It's a very generalist, sap sucking insect that might cause damage to water culture and agriculture it establishes in Australia. There are about 600 known Australian native species of stink bugs, but there may be thousands of undescribed species.

The critical thing about invasive species is that they don't have any natural enemies here, so they may be able to do a lot more damage than the native ones that are naturally control.

The present project is funded by the Department of Agriculture, Water and the Environment, through the Biosecurity Innovation Programme. We are creating an image recognition model that will allows inspectors and biosecurity officers to distinguish the Brown Marmorated Stink Bug and other potential pest species from close related and similar native species. That will allow them to make decisions about cleaning cargo, for example, about eradication actions and will ultimately safeguard the Australian economy and agriculture.

We have produced this AI model, but to really make it usable to people, we are putting it into a smartphone that the Department of Agriculture can then use at the border, at ports to identify specimens ‘in the field’, so to say.

I myself am an optimist and actually the first model that we produced and that we put into this prototype app here was a wheat seed identification model because I'd come out of wheat seed identification project for the department.

And when we showed this to partners at the Department of Agriculture, Water and the Environment, it turned out that, yes, the principle of this identification app was great. But the actual, really concrete problem at the moment was the stink bugs, so that is where we then turned our attention next. But ultimately an app like this can have all manner of identification models. We could have one in the future for moths, we could have one for wheat seeds again, we could have one for rust fungi potentially. Anything that lends itself to being visually recognised by the expert can also then be visually recognised by the artificial intelligence.