Video Transcript

CSIRO: Using Artificial Intelligence To Reduce Plastic Waste.

 How big is the plastic pollution problem globally? We anticipate the plastic production will double by 2040. We're losing thousands of metric tons of plastic pollution to the ocean. 

Each and every year, we find everything from cigarette butts, take away containers, plastic cutlery, lids, bottle caps in Sydney Harbour alone. It's estimated that 3,500 cubic meters of plastic pollution are lost to the harbour each and every year.

To put that into context., 3,500 meters of plastic going into Sydney Harbour is equivalent to around 44,000 Wheelie bins full of trash each and every year dumped into our waterways. 

We find plastic being eaten by seabirds, marine mammals, turtles, and almost every species. We have found more than 700 species of marine wildlife have interacted with plastic typically by mistakenly eating it or getting tangled up in it.

So, what can we do about this massive problem? 

Well, at CSIRO, we're working using smart sensor technology, cameras and myriad of other tools to help solve plastic pollution, preferably before it gets out into the environment. 

Using technology is a smart, safe, cost effective way for us to fundamentally change the dial on plastic pollution in Australia and beyond.

One thing that we're doing is quantifying and identifying floating plastic in our rivers along our waterways. 

The other main technology that we're using is smart sensor technology that communicates over the wireless network so that we can understand when and where our stormwater assets are full. I am full and need to be emptied.

Well, after working on the problem more than 15 years, I realised the importance of stormwater drains in our stormwater assets in helping us to resolve this issue. 

Our team at CSIRO is using computer vision and machine learning algorithms by which we're training a computer to recognise the floating objects.

We use people to train the computer model, and that allows us then to process thousands to millions of images much more quickly and much more cost effectively than we could by having a human look at each of millions of images. 

We can put cameras out at bridges, on rivers and waterways around the country, around the world, and use the images that we get back to better estimate and understand the types and amounts of floating plastic that are coming through our waterways. 

We're able to use inexpensive wildlife cameras. We set a time lapse interval to take these images of trash floating down our waterways. Our team right now is focusing on a catalogue of images that include over 14,000 different objects that are categorised into around 35 categories.

Bottle caps, lids, cardboard, cutlery, containers, polystyrene, all these different categories so that we can identify what are those particularly problematic items that are lost into the environment most regularly. 

If we as a global community, change our relationship with plastic, if we think about what we use and how we use it, if we identify those particularly problematic items that are lost into the environment most often, and we know which of those results in the most harm, we can target interventions, we can target solutions much more effectively, much more efficiently.