How the Wildlife Conservation Society uses AWS to accelerate coral reef monitoring worldwide

The Wildlife Conservation Society (WCS) envisions a world where information quickly and effectively moves from the hands of scientists to communities and governments, enabling rapid action to protect coral reefs. Through their MERMAID (Marine Ecological Research Management AID) platform, WCS creates transparency around coral reef health and empowers people and institutions with the data they need to take action to save reefs.

With support from the AWS Imagine Grant that provides both cash and Amazon Web Services (AWS) credit funding to registered nonprofit organizations who are using cloud technology to accelerate their missions, MERMAID is evolving to integrate machine learning workflows. This innovation allows any diver to take reef photos, have them automatically analyzed, and contribute to open global datasets. The project aims to double the amount of reef data in MERMAID, accelerate policy action, and move closer to the goal of monitoring 100 percent of the world’s reefs by 2030.

The urgent challenge facing coral reefs

Coral reefs face an unprecedented crisis. Half have already been lost, and up to 90 percent could disappear by mid-century without urgent action. Yet reefs that are more resilient to climate change still exist, and protecting them could safeguard biodiversity and the well-being of one billion people.

The problem is that less than 12 percent of reefs are monitored today, leaving governments without the information they need to prioritize climate-resilient reefs in their conservation commitments. Before MERMAID researchers often spent weeks or months manually classifying thousands of photo quadrats or navigating multiple platforms to process and clean their data. This created significant reporting delays and prevented scientists from focusing on critical conservation decisions.

Since launching in 2018, over 2,000 scientists in 44 countries have used MERMAID to upload more than 50,000 surveys and 8.5 million observations. However, the manual nature of image processing remained a bottleneck that limited the platform’s potential to scale reef monitoring efforts globally.

Beneath the waves of Fiji, the silent work of conservation unfolds as divers diligently assess the vibrant, yet vulnerable, coral reefs. Photo credit: Tom Vierus

Building AI-powered reef monitoring with AWS

To implement MERMAID, WCS built directly on AWS cloud-native infrastructure to integrate image classification into MERMAID’s existing data workflows. Their comprehensive AWS stack includes:

• Amazon Simple Storage Service (S3) to host user-uploaded reef images and confirmed annotations
• Amazon Elastic Container Service (ECS) to power the API and asynchronous worker tasks that handle survey data processing and classification requests
• Amazon Relational Database Service (RDS) PostgreSQL as MERMAID’s central database
• Amazon SageMaker with MLFlow for training, deploying, and comparing machine learning models
• Amazon Simple Queue Service (SQS) to queue image classification tasks
• AWS Lambda to trigger serverless functions for processing and integration

The team developed their proof-of-concept and beta model using CoralNet’s Pyspacer framework, which they are now expanding into a fully managed workflow using Amazon SageMaker. This approach allows them to retrain models on demand as new images and annotations are uploaded, and to host those models using SageMaker endpoints for consumption by the MERMAID API and Collect app.

The AWS Imagine Grant resources supported the scaling of this infrastructure. The combination of cash funding and AWS credits helped WCS to build a scalable, secure, and cost-efficient backbone that supports thousands of scientists and citizen divers worldwide uploading, classifying, and sharing coral reef monitoring data in near real time.

Transforming coral reef research worldwide

When WCS launched MERMAID AI on World Oceans Day in June 2025, it marked the first time that benthic photo quadrat analysis could be done directly inside MERMAID. Instead of importing photo annotations created in other platforms, users can now drag, drop, and use AI to classify photos seamlessly alongside their fish, bleaching, and habitat data.

Powered by a shared model trained on more than 500,000 public CoralNet images in partnership with the University of San Diego and Scripps Institute of Oceanography, MERMAID AI can currently identify 54 attributes from broad benthic groups like hard coral and macroalgae to 37 coral genera and their growth forms. Early testing shows promising accuracy (0.82) and, with user verification built in, the model continues to improve over time.

The quantifiable results demonstrate significant impact. Since launch, users have uploaded over 2,200 images and more than 55,000 annotations directly into MERMAID. Over 3,000 MERMAID users in 52 countries now have access to the same AI model, creating consistency across projects and reducing barriers for smaller teams with limited technical capacity.

The qualitative improvements are equally impressive. WCS field teams in Mozambique shared that what used to take days of back-and-forth data wrangling is now streamlined into a single step, freeing up valuable time to focus on science and management decisions rather than data processing. Scientists are processing images faster, reducing friction in their workflows, and saving hours of manual data entry.

The human-in-the-loop design builds trust by keeping users in control while still benefiting from automation. Users report feeling empowered knowing their data contributes to a shared open platform that will accelerate global reef science while protecting user privacy.

Looking ahead, WCS will track adoption metrics, images processed, time saved, and model performance as user verifications retrain the system. They will also measure bigger-picture outcomes: increases in the share of reefs monitored with photo quadrats, growth of the open image repository, and policy applications where MERMAID AI data informs decisions on climate-resilient reef protection.

Using MERMAID, a scientist in the Philippines contributes real-time data from a resilient reef to the platform’s cloud-based monitoring system. Photo credit: Emily Darling

Lessons learned for nonprofit technology projects

The WCS team offers valuable advice for other nonprofits implementing technology projects.

First, start with your users. New technology is only valuable if it solves real problems. For MERMAID AI, they listened closely to scientists who told them that image processing was slowing their work. By grounding development in user feedback, they built a tool people were eager to adopt.

Second, don’t try to reinvent the wheel. Partnering with existing open-source platforms like CoralNet saved years of effort and allowed them to focus on integration and scaling. Nonprofits often feel pressure to build everything, but collaboration and open science can unlock far greater impact.

Finally, invest early in cloud architecture and security. The team underestimated the time required to design scalable AWS workflows and to build trust with users about how their images would be stored and shared. Working closely with AWS solutions architects proved to be invaluable, and they wish they had engaged them from the start.

This is just the beginning for MERMAID AI. With the community’s contributions, the platform is poised to help close the global coral reef data gap and move closer to the vision of 100 percent reef monitoring by 2030, protecting these critical ecosystems for future generations.

How you can support MERMAID AI and WCS

If you’re a scientist, diver, or student working on coral reefs, WCS invites you to join the community by trying MERMAID and sharing your feedback—every image and annotation helps make the model stronger for everyone. You can learn more and create a free account on this webpage.

If you’re a funder, policymaker, or partner organization, the WCS team would love to connect to discuss how MERMAID can support your work to protect climate-resilient coral reefs.

And if you’re part of the tech or corporate community, the WCS team invites you to partner with them as part of your ESG commitments. By sponsoring MERMAID, you’ll help bring cutting-edge cloud and AI tools to conservation—and directly contribute to protecting the world’s coral reefs and the communities that depend on them.