Northern Territory nets advantage with cloud and AI: Automates fish census and accelerates scientists’ ability to protect the environment

Using a GoPro, researchers at NT Fisheries begin the process of assessing critical fish species.

SYDNEY/DARWIN – November 30, 2018 – Australia’s Northern Territory has deployed a Microsoft AI solution to help monitor and manage marine health by rapidly analysing underwater video captured around Darwin Harbour.

An open source solution, developed in association with Microsoft, and available on GitHub, the AI platform automates the laborious process of counting local fish stocks by progressively learning to identify different varieties of fish. Built using Azure Cognitive Services, it has global conservation potential as it can be trained to spot an array of different animal and fish species, using techniques similar to those used for facial recognition in social media.

The environmental importance of the Department of Primary Industry and Resources (DPIR) Fisheries Department’s work cannot be overstated. The Northern Territory boasts a pristine marine environment with healthy populations of species such as sharks and sawfish. Its fish and marine resources support customary use, commercial and recreational fishing as well as tourism.

The value of goods and services produced by the Territory’s primary industries and fisheries reaches over half a billion dollars each year.

Unable to enter the water directly because of risks posed by saltwater crocodiles and sharks, DPIR scientists attached underwater cameras to buoys in protected reefs off Darwin. They then spent tens of hours watching footage in order to spot and count fish. Now the AI system analyses hours of video in minutes, freeing up scientists for more valuable ecosystem sustainable management work.

fisheries gif
Microsoft AI helps identify fish species in the murky waters of Darwin Harbour

The AI solution has been developed at a time when global fish stocks face greater pressure than ever before. Fish play an increasingly important role in feeding the global population. According to the United Nations’ 2018 fisheries and aquaculture report,[1] between 1961 and 2016, the average annual increase in global food fish consumption (3.2 per cent) outpaced population growth (1.6 per cent) and exceeded the increase in consumption of meat from all land-based animals combined (2.8 per cent).The fact that the solution is open source creates potential for similar platforms to be deployed in different settings around the world, to support important scientific endeavours that will benefit the earth’s environment and humanity.

Azure speeds development

The creation of the AI solution was completed within a short timeframe. Using Microsoft Azure AI services, the first iteration of the system was up and running in a month. The solution was widely deployed within six months, and its identification powers have been progressively enhanced using machine learning ever since. The AI system is now able to identify a fish in a video with 95 to 99 per cent accuracy.

DPIR fisheries scientist, Dr Shane Penny, said that two particular species have been of early interest to the team; golden snapper and black jewfish.

Steve van Bodegraven, Machine Learning Engineer at Microsoft and Dr Shane Penny, Fisheries Research Scientist at NT Fisheries review the identified fish species using the AI solution.
Microsoft Machine Learning Engineer Steve Van Bodegraven with Dr Shane Penny. 

“These are two commercially and recreationally important species in the Northern Territory, but research has proven that they had been overfished around the greater Darwin area,” said Penny.

Part of the management plan for those species was to put in place a number of protection zones, with underwater video installed to monitor numbers in the area.

Lee Hickin, National Technology Officer, Microsoft Australia said: “Cloud computing and AI are combining to support scientists to gain a deeper understanding of fish populations. Freed from the mundane aspects of counting and identifying fish, scientists can instead take the insights from the AI solution and focus on making informed decisions that have significant environmental and economic impacts.”

Broad applications and efficiency dividends

Further potential applications across the NT are already being considered, including monitoring feral fish in the freshwater systems of the NT and cattle movements across the Territory.

DPIR chief information officer, Rowan Dollar, is keen to explore the potential regulatory applications of the technology, for example keeping an eye on the commercial catch.

“We could look into setting up a camera on a trawler that’s out at sea and doing on-the-fly identification of the catch, so we can start measuring by-catch. We can start being able to identify that in real-time, to help better manage those fisheries.”

For a Government department which runs lean, the efficiency dividend of these intelligent cloud-based systems is important.

“It’s important for every government jurisdiction, regardless of who you are or where you are, to be using technology to gather and analyse data. It’ll help you be more efficient and give better value back to your stakeholders,” said Dollar.

[1] http://www.fao.org/3/i9540en/I9540EN.pdf

Related Posts