REEF SUPPORT x SOA: AI FOR CORAL BLEACHING 
HACKATHON IN SAUSALITO, CALIFORNIA


17/03/2023
Written by Marcel Kempers

Reef Support had the privilege of being the challenge owner of the AI for Coral Bleaching hackathon with Sustainable Ocean Alliance in Sausalito, California. The event brought together some of the brightest minds in the marine conservation and technology fields to develop innovative solutions to one of the biggest threats facing our oceans today: coral bleaching.

The Reef Support team consisted of marine biologists, data scientists, creative consultants, business analysts, technologists, and designers. Our diverse skill sets allowed us to approach the challenge from multiple angles and develop a comprehensive solution to the problem.

Our team consisted of Hannah Lochan, a Marine Biologist, Violet Sinnarkar, a Data Scientist, Tosha Grantham, a Creative Consultant Citizen Scientist, Alex Merkle-Raymond, a Business Analyst and Research Administrator, Xander Peterson, a Technologist and Data Visualizer, and Erika Woolsey, PhD, a Marine Biologist and Designer.

Hackthon Experience

During the hackathon, the team worked tirelessly to research and develop a business plan around reef monitoring and using artificial intelligence. Our team members from small islands brought valuable perspectives on the local challenges and solutions needed for effective coral reef conservation.

The first step in the team's process was to gather data on the current state of coral reefs and identify available datasets and the use case for AI software for coral bleaching detection. We analyzed data from various sources including online databases, field studies, and citizen science initiatives. We are lucky that our team is made up of smart marine experts so that they could immediately tell which data is of quality or not, because this is what some of them do on a daily basis! Take Hannah for example, who works in Trinidad en Tobago.

The next step was to develop an AI model that could accurately predict coral health and bleaching events. Our data scientists created a machine learning model that was trained on a small dataset of images of healthy and bleached corals. The model was able to detect changes in coral color and morphology that indicated bleaching, allowing us to predict bleaching events before they happened. The extremely talented Violet, Hannah and Xander came together for this part.

To make the model more accessible to researchers and citizen scientists, our technologists developed a user-friendly frontend that could be used to upload images of coral and receive a prediction of the coral's health. This allowed anyone with a smartphone to contribute to our database of coral reef images, which could be used to train and refine the AI model.

The team also developed a business plan and pitch that outlined how our AI solution could be used to support coral reef conservation efforts around the world. Our plan included partnerships with local NGOs and government agencies, as well as a subscription-based model for researchers and conservationists to access our AI tools.

During the hackathon, we worked closely with our team members and mentors to create an AI model that could detect coral bleaching and live coral cover using k-means segmentation. We validated the model by hand-annotating about 100 pictures, which helped us refine and improve our results.Despite not winning the final prize, we are proud of what we accomplished and the progress we made in such a short amount of time. We are excited to continue building on this baseline AI model and explore its potential applications for ongoing marine research and coral reef monitoring practices.

Overall, the hackathon was a fantastic opportunity to collaborate with like-minded individuals, learn new skills, and work towards a common goal of protecting our oceans and coral reefs. We are grateful for this experience and look forward to staying connected with the Sustainable Ocean Alliance community.

The Challenge

For this challenge, we set out to focus on AI for Coral Bleaching.

The Nature Conservancy, Sustainable Ocean Alliance, and Reef Support came together for the AI for Coral Bleaching Hackathon, a collaborative effort to explore the potential of AI in coral reef conservation. The Nature Conservancy, a global environmental organization, brought their expertise in coral reef conservation and management to the table. Sustainable Ocean Alliance, a youth-led organization focused on ocean sustainability, provided a platform for young innovators to come up with solutions to pressing ocean issues. Reef Support, a startup dedicated to using AI and machine learning for coral reef conservation, was the challenge owner of the hackathon.

The use of AI tools for automated processing of coral data is an important development for the field of coral conservation and management. With the current state of our oceans, coral reefs are facing unprecedented threats such as coral bleaching, ocean acidification, and overfishing. As a result, the need for efficient and effective monitoring and management strategies has become more pressing than ever before.

AI tools for coral bleaching can play a crucial role in this process by providing automated and accurate analysis of coral data. This can help identify the extent of coral bleaching in a timely manner and allow for early intervention to mitigate its effects. The use of AI can also help to streamline the process of data collection and analysis, which is crucial for large-scale monitoring efforts.

In addition, AI tools can assist with the identification of key stressors affecting coral reefs. By analyzing a variety of environmental and biological data, AI algorithms can identify patterns and correlations that may not be immediately apparent to human researchers. This information can then be used to inform management decisions and guide conservation efforts.

Final word

The hackathon experience was incredibly rewarding for the Reef Support team. It was inspiring to work alongside so many talented individuals who were passionate about using technology to protect our oceans. We were able to develop a comprehensive solution that could have a real impact on the future of coral reef conservation.

Looking back on the experience, we feel grateful for the opportunity to have participated in such a meaningful event. We learned a great deal about AI and its potential applications in marine conservation. We also gained valuable experience in teamwork, innovation, and problem-solving.

We also want to take a moment to extend a heartfelt thank you to Tamaki Bieri from Nature Conservancy for her invaluable guidance and support during our recent hackathon with Sustainable Ocean Alliance. Tamaki's expertise and insights were instrumental in helping our team at Reef Support develop our AI tool for coral bleaching detection.

Moving forward, we plan to continue refining our AI model and expanding our database of coral reef images. We also hope to establish partnerships with organizations and governments around the world to implement our solution on a larger scale. By working together, we can protect the world's coral reefs and ensure that future generations can enjoy the beauty and biodiversity of these incredible ecosystems.

A Hackathon Baby!

Last by but least, we would like to give the greatest congratulations to Xander Peterson and Erika Woolsey on your exciting news! As a team, we were thrilled to hear the news and we couldn't be happier for you both. Thank you for sharing this wonderful news with us and we wish you all the best in this new chapter of your lives. Your contributions to our team and to the field of marine science are greatly appreciated and we can't wait to see what the future holds for you both. Congrats again!

If you're interested, some Methods and Results the team accomplished

During the AI for Coral Bleaching Hackathon organized by the Sustainable Ocean Alliance, the Reef Support team used the k-means segmentation technique to detect live coral cover and coral bleaching extent. This was a critical step in creating an accurate AI model to predict coral bleaching events and identify healthy and unhealthy coral reefs.

K-means segmentation is a popular unsupervised machine learning technique used to cluster similar data points. In the context of coral reefs, this technique was used to segment images into different clusters based on color and texture. By identifying clusters that represented live coral cover and those that represented bleached coral, the Reef Support team was able to create a model that accurately predicted the extent of coral bleaching in an image.

To validate the accuracy of the model, the team manually examined and labeled approximately 100 images to determine the actual percentage of live coral cover and bleached coral in each image. The results of the manual validation were compared to the predictions made by the AI model, and the model was found to be accurate in detecting live coral cover and coral bleaching extent.

For more information about Open Coral AI developments, please visit the page below

This project is a project very close to our hearts at Reef Support. It isn’t only about helping coastal communities in their effort to monitor and restore the reefs but also about fostering wide-spread marine knowledge!

If like us you are fond of our oceans and want to save these beautiful creatures which are so important to our ecosystem, we would like to give you the opportunity to adopt your own coral spider. And what is even cooler is that they will tag your name personally on it and give you a monthly report of its coral growth! Visit the page below to find out more on how to participate.

Reef Support Wave
OPEN CORAL AI
Reef Support Wave