The Great Barrier Reef is a globally outstanding and significant entity. The entire ecosystem was inscribed as a World Heritage site in 1981. Located in Australia, the reef covers an area of 348000 Square kilometers and extends across a contiguous latitudinal range of 14 degrees. According to UNESCO, the Great Barrier Reef (GBR) includes extensive cross-shelf diversity, stretching from the low water mark along the mainland coast to 250 kilometers offshore. 

According to a 2012 study on the world’s most extensive time series data on reef conditions, the GBR has lost more than 50% of its initial coral cover. Factors such as climate change have caused the reef to suffer from various “bleaching events”. Bleaching is when the coral expels the algae living in its tissue due to the higher warmth in the water. 

The bleaching of coral reefs does not result in their death. They can even recover from their condition, but the process takes up to 12 years. The conservation efforts of GBR are ongoing. However, the size of the reef and the land it occupies means that these efforts can only be focused on small sections. The researchers studying the reef have only been able to monitor around five to ten per cent of the 3000 individual reefs.  

Capturing images 

The volunteers studying the reef collect images of the reef using dive boats, tourism vessels, watches, fishing charters and tug boats. These images are used to understand the health of reefs and to identify the location where conservation efforts must be focused. As part of the project, researchers from the University of Queensland, Dell Technologies and other organizations came together to use AI and crowdsourcing to understand the reef better. Two Great reef censuses have already been conducted. 

In the second census, the researchers crowdsourced over 40000 images of coral in GBF. The major challenge in handling the images was tagging and identifying coral species within the images to see which ones needed to be studied by marine biologists. 

AI for biodiversity 

The process of tagging and identifying the images is time-consuming. Dell Technologies developed a Deep Learning model to aid the process. The trained model analyzes every pixel in an image to classify the reel and coral infrastructure. 

Indian Express had reported the statement of Aruna Kolluru, Chief Technologist, Emerging Technology at Dell, that the Deep Learning model identifies the part of the image containing coral. It then classifies them if they belong to selected species of coal. This data, along with the location data of the image, can provide researchers with a snapshot of what kind of coral exists.  

The AI model can take on the repetitive and tedious identification and classification of the coral. This allows more time for the volunteers for data collection. Based on the classification done by the model, marine biology experts, if necessary, can analyze the data or output. Presently, the time taken by the model for analysis is one minute.  

The model will be developed to analyze images more accurately and also to identify more species of coral. As the model develops further, researchers look forward to protecting other reef systems globally. 

Sources: 

UNESCO 

Indian Express 

PNAS