CAW01 - AI for Coral Reefs

Reef support is a Dutch start-up which focuses on building software to accelerate the transition from data to decisions to protect coral reefs. Their mission is to take practical action and support the blue economy with imagery and artificial intelligence. They aim to accomplish it by building robust research and predictive maintenance tools for research institutions, reef managers and other coastal activities – from small conservation efforts to large-scale coastal developments.

Due to human impact and the rise in climate change, coral reefs and the health of its inhabitants are in a dire situation. Reef Support aims to aid this situation and bring relief to an already hurt ecosystem. The proposed challenge is based on computer vision and image recognition.

The objective of Coral AI’s Hackathon challenge is to use the image data alongside the tabular data (especially the annotation table) to define certain metrics:

The first level of analysis would be to determine the functional group of corals in the photo-quadrats by using the pixel-level data in the annotation table. Then, the second level of analysis would be to add granularity to the analysis and determine the labels of this pixel-level data. Finally, the third level of analysis would be to use image segmentation to determine coral boundaries on the photo-quadrat images and then use colour analysis to make out the percentage of bleaching corals in each image. This final level will allow us to obtain a deeper understanding on the reef’s health and resiliency.

Computer vision expert
Marine biologist
Data scientist