How computer vision can protect coral reefs

The Australian Institute of Marine Science (Aims) has teamed up with Accenture to monitor the health of coral reefs in Australia and other marine ecosystems using computer vision.

Using an Aims database of 6,000 images from six different ocean regions, the technology helps to automate the analysis of coral reef images, so that researchers can understand the response of specific coral species to stressful scenarios such as bleaching events, among other things.

These images are currently analysed and labelled manually, limiting the scale to which images can be analysed and ultimately slowing down preservation efforts, said Richard McNiff, rapid innovation director at Accenture’s The Dock, a research and development facility in Dublin, Ireland.

“Using computer vision to automate this process will provide the teams with the means to label images much quicker and with more detail,” McNiff told Computer Weekly. “This will also allow for more advanced image analysis and at scale.”

McNiff said the project was inspired by the Chasing Coral documentary on Netflix. “We were looking at ways to use artificial intelligence to help the environment. Ocean conservation was raised as an area of personal interest within the team,” he said.

A multidisciplinary team from Accenture then reached out to Aims on the partnership, and subsequently developed three different computer vision approaches to help Aims understand alternative approaches to coral monitoring.

McNiff said: “The findings from this initial work were very promising, with one particular experiment – a graph-based approach – showing real potential. Aims were impressed and in conjunction with The Dock, we explored other areas that could result in a significant step change in terms of the applications of machine learning, cloud computing and data analytics for coral reef monitoring and conservation.  

“This technology, combined with clever use of interface design, enables us to understand what the data is saying, how healthy the reefs are, how healthy the coral is and which species of coral we should be replanting in order to conserve the reefs into the future.”

McNiff said the computer vision system was designed with the human in the loop, and so any suggestions made by the computer vision algorithm can be confirmed by a human. “This helps minimise the progression of false positives through the learning process,” he added.

Coral reefs worldwide are under threat. The condition of many coral reefs around the world has declined from the cumulative effects of global and local challenges, through extreme weather events like marine heatwaves, tropical storms, ocean acidification and chronic pollution.

With predictions that about 75% of coral reef globally may be severely threatened, not only are the habitats of a quarter of all marine life at risk, but so too are the food sources, incomes and livelihoods that are central to the economic future for one billion people globally.

“Australia is an economically well-developed nation hosting coral reefs, including the iconic Great Barrier Reef, and the purposeful development of these technologies will benefit nations across the world towards gaining efficiency in coral reef monitoring and fostering integrative efforts to better manage our reef systems,” said Manuel Gonzalez Rivero, senior research scientist at Aims.

“Through our work with Accenture, we seek to bring innovative and transformative change to the way we monitor our reefs. Such a technological and efficient solution will provide scientists, environmental managers and decision makers with the knowledge that contributes to the sustainable productivity of marine-based industries and helps protect coral reef ecosystems globally,” Rivero added.

Accenture’s collaboration with Aims follows an earlier project that it had undertaken with Intel and the Sulubaaï Environmental Foundation in the Philippines to analyse coral reef resiliency using underwater cameras that track the movement of fish around the coral reef as an indicator of the health of the reef.

Data from the cameras was transmitted to the cloud, where artificial intelligence was applied to count and classify different types of marine life, giving researchers an overall sense of how the reef was faring.

“Our project with Aims looks at the species of coral itself, rather than focusing on the fish – understanding the health of the coral, understanding resilient species and planning intervention strategies to sustain and/or replant the reef,” McNiff said.

McNiff noted that both projects approach coral reef conservation from different but complementary angles and can be part of an overall conservation effort in future.