Despite having more coral species than the Great Barrier Reef, Darwin’s waters are under threat from introduced species, overfishing and climate change.
In an effort to better manage and control fish populations in the muddy, crocodile-infested waters of Darwin Harbor, the Government of the Northern Territory has automated the process by feeding frames from cameras installed on buoys to an artificial intelligence platform.
Scientists from the Department of Primary Industry and Resources (DPIR) turned to GitHub and Microsoft Azure Cognitive Services to develop an open source AI solution, spending dozens of hours identifying the fish that floated in front of the cameras to train the machine learning program Platform.
The first iteration of the system was launched and launched within a month, and over the next six months of development and deployment, its identification capabilities gradually expanded to such an extent that AI could now identify fish with an accuracy of 95 to 99 percent.
Fish identification speed is also accelerating to such an extent that AI can analyze video hours in minutes, freeing DPIR scientists for more valuable work on ecosystem management.
DPIR fishing scientist Dr. Shane Penney said the system allows them to monitor overcatching species like the golden bass and the black Jew, and also better adapt conservation efforts.
“These are two commercially and recreationally important species in the Northern Territory, but research has proven that they were outsmarted around most of Darwin,” said Penny.
DPIR Chief Information Officer Rowan Dollar is already seeking to expand the program based on its early success, including monitoring of wild fish in freshwater systems and equipping commercial boats with cameras.
“We could study the camera setup on a trawler that is at sea and identify the catch on the fly so that we can begin to measure bycatch. We can begin to understand that in real time, to help better manage these fisheries, ”said Dollar.
Other fisheries around the world could also adopt a system that was uploaded to GitHub to manage fishery stocks under various conditions.
Only in the Northern Territory, the value produced by primary industries, including fishing, reaches more than half a billion dollars a year.
It is expected that the demand for marine NT resources will grow in line with global trends, when fish consumption increased by 3.2 percent per year in accordance with the 2018 United Nations report.
Other government conservation agencies are also looking for computer vision solutions to manage sensitive ecosystems, including the Queensland Department of Agriculture and Fisheries, which tracks invasive plant species in conjunction with QUT.