Machine learning with Microsoft

We are utilising Microsoft machine learning tools and capabilities to help tackle illegal, unreported and unregulated (IUU) fishing.

An explosion from fish blasting in the ocean with a small boat near by

Utilising Microsoft technologies to process data from camera and underwater recording devices will improve the detection of illegal fishing,

The tools will expediate data processing by analysing images gathered from high resolution cameras and audio from sensor systems in waters off Indonesia and the Great Barrier Reef. This includes assisting with fishing management in Australian protected areas and identifying boats and their behaviour using underwater explosions to stun and harvest fish as part of illegal ‘blast fishing’ in Indonesia.

High resolution cameras are a cost-effective method for marine monitoring and surveillance. Audio sensors called hydrophones can record sound underwater within a range of 10 kilometres and 30 metres below the surface. They can send a real-time notification to law enforcement should IUU fishing be detected.

AI technologies accelerate IUU fishing research

IUU fishing is a major challenge for sustainable fisheries management, particularly in the developing world. Applying Microsoft machine learning will enable us to process large sets of data in real time to detect IUU fishing activities quickly and efficiently.

Existing methods include ship tracking systems and water user data. However, it is difficult to transform the data into useful information. It is also labour intensive.

Indonesia uses a vessel GPS tracking system. However, there are roughly half a million fishing vessels, making inspecting individual vessels nearly impossible.

Our approach can capture the type of boat, boat features, boat travel speed, idling, diving and if any fishing bombs are used. Further development will alert authorities to suspicious fishing behavior automatically, to assist with investigation.

Supporting fisheries management

Our collaboration with Microsoft builds on our existing research and accelerates the processing of IUU fishing data in a cost-efficient way.

Our trained algorithms for image recognition will help to detect IUU fishing patterns including the real time processing of data. It also improves privacy, as large amounts of data do not have to be stored.

This information will be critical for fisheries agencies to manage IUU fishing in Australian and international water. It will also help to ensure food security and livelihoods for the world’s growing population.