QUT to use AI and ML in Gold Coast future waterways project
The university will help the Gold Coast Waterways Authority understand how artificial intelligence and machine learning technologies can be used to identify vessel type and determine the number of users of its waterways network.
The Queensland University of Technology (QUT) will be assisting the Gold Coast Waterways Authority (GCWA) in its smart camera trial that aims to build a detailed picture of activity on the city's waterways network.
In combination with information on marine incidents and weather conditions, QUT said it will establish usage trends and provide insight into future patterns of waterways use and pressure points.
A second component of the project is exploring the feasibility of using the same camera technology to develop a more accurate way of measuring vessel speed on the water, the university added.
"We'll be using the cameras to create a more complete picture of who is using the waterways, where they're going, what type of vessel or watercraft they're travelling in, and how they're interacting with other users and the environment," GCWA CEO Hal Morris said.
"This information hasn't been collected before to this level of detail on the Gold Coast. It's important because to successfully plan for the future we need to understand the impacts population growth and rising boat ownership are having so that we can plan for these changes, protect the environment, and ensure locals and visitors continue to enjoy safe access to our beautiful waterways city."
Cameras have begun rolling out at 20 locations around the Coomera River and Southern Broadwater, with GCWA saying several are "ready to start snapping away" on the Australia Day long weekend, which Morris said is traditionally one of the busiest on the waterways.
The cameras will take continuous photographs at their locations at all times of the week and in all weather conditions.
QUT will then use image analysis to automatically process the photographs and advanced machine learning methods to understand what features of the images can be used to identify vessel type and determine the number of users.
"From this, we will develop a statistical model that will incorporate additional information about, for example, weather and marine incidents, to provide an indicator of future patterns of waterways use," QUT project manager, associate professor James McGree said.
The project will also investigate whether computers can be trained to recognise vessel registration numbers to help with the identification of speeding vessels, QUT said.