General FAQs
MERMAID is a free cloud-based platform to collect, visualize, and analyze data on coral reef ecosystems. MERMAID helps scientists, reef managers, communities, and governments quickly assess and track changes on coral reefs to monitor existing conservation efforts or identify new areas as conservation priorities.
MERMAID is actively developed by the 501(c)(3) nonprofit Wildlife Conservation Society (WCS) and the geospatial consultancy Sparkgeo. It is continuously updated based on the knowledge, input, and needs of experts and stakeholders working directly with coral reef monitoring. MERMAID is an open-source application and all development code is available on GitHub.
MERMAID is a free, open platform for anyone collecting or using coral reef data. Researchers, NGOs, governments, and community groups worldwide use MERMAID to record, analyze, and visualize coral reef status and trends. More than 3,000 users from 180 organizations are already part of the MERMAID community — explore their work on MERMAID Explore.
MERMAID supports benthic surveys (line intercept transect, point intercept transect, photo quadrat transect), fish surveys (belt transect), bleaching surveys (rapid assessment using quadrats), and habitat complexity surveys (visual scoring). We are currently implementing a macroinvertebrate survey.
Have another survey you'd like to see us add? Please reach out.
Yes. MERMAID fully supports offline data collection, so you can work in the field without any internet connection. Before heading offline, make sure to set up your project, add users, and test the offline workflow.
Learn how to prepare for offline use in our documentation here. You can also download a PDF version of our full documentation so you have helpful information for troubleshooting and answering questions while working offline.
Yes. MERMAID includes an integrated AI image classification model, currently in beta, that automatically identifies benthic groups and coral genera from uploaded photo quadrats. This provides an easy entry point to AI-powered monitoring with a simple and accessible workflow integrated with your other MERMAID workflows.
MERMAID ensures that your data are safely stored, backed up and encrypted. Our cloud infrastructure is hosted by Amazon Web Services (AWS), one of the world's most robust cloud infrastructures and maintains the highest standards of data protection and privacy.
Read more in our Terms of Service.
You and your team control all data access and sharing in MERMAID. Only users with the Admin role can set or change a project’s data-sharing policy, and raw data are visible only to users within that project unless you choose to share them more broadly.
MERMAID includes three user roles—Admin, Collector, and Read-only—each with different permission levels.
Learn more about user roles here.
No. You always retain full ownership and control of your data in MERMAID, and you decide whether—and how—you want to share it with the broader community. If you prefer not to share any raw observations or summaries, you can select the Private data sharing policy.
That said, we encourage users to share data whenever feasible, as shared information strengthens reef science and conservation efforts globally. To support different comfort levels and project needs, MERMAID offers flexible data sharing policies—from fully open access (Public), to summary-level sharing only (Public Summary), to completely restricted (Private)—so you can choose the policy that best fits your scenario.
Explore how to choose the right data sharing policy for your project here.
MERMAID enhances data standardization and validation to ensure consistency in data analysis and reporting. After users conduct surveys and enter observations into MERMAID, a series of QA/QC checks are performed automatically to catch common errors or typos. These checks are then validated by users before data are submitted, ensuring data reliability.
There are many stories of MERMAID's impact, we hope yours can be next!
In the Persian Gulf, Dr. Jeneen Hadj-Hammou and her team at NYU Abu Dhabi used MERMAID to compile historical coral reef data from seven countries, enhancing collaboration, improving data accuracy through QA/QC processes, and saving time with automatic fish biomass calculations.
Rare's Fish Forever program transformed 'a nightmare' of data management by implementing MERMAID, leading to better conservation outcomes and community engagement.
In Mozambique, WCS has shared MERMAID with national coral reef partners to visualize coral reef metrics quickly on MERMAID Explore, driving effective conservation efforts and inspiring broader public engagement.
Yes. MERMAID will always remain free to use. As an initiative of the non-profit Wildlife Conservation Society, our mission is to keep MERMAID and its full suite of tools open access for everyone. While some optional services or partnership arrangements may involve costs, the core platform—including data collection, visualization, and AI-powered image classification—will always be free for all users.
MERMAID AI FAQs
MERMAID AI is built directly into the MERMAID Collect app, so anyone gathering coral reef data can use the AI model instantly—no technical setup required. Just drag and drop your photo quadrats, and MERMAID automatically identifies major benthic groups and coral genera as part of your normal workflow. Your results appear alongside all other ecological data in MERMAID Explore.
MERMAID AI was developed in partnership with CoralNet, and our classifier is powered by their pyspacer deep-learning engine. While platforms like CoralNet or ReefCloud are ideal for users who want to build, train, and manage their own custom models, MERMAID provides a pre-trained, ready-to-use model that works out of the box. There’s no need to create labels, train models, or run validation steps.
The result: AI-powered image analysis with zero technical barriers—making high-quality coral-reef monitoring accessible to scientists, managers, and community programs of all experience levels.
Accuracy varies depending on the benthic group or coral genus being identified—some are much easier for the model to recognize than others. Current accuracy (measured using the F1 score) ranges from ~0.3 for harder-to-distinguish groups like Millepora to ~0.9 for more visually distinctive groups like Montastraea.
As more images and expert-verified annotations are added to MERMAID, future model versions will become more accurate across a wider range of groups.
MERMAID uses the F1 score to measure accuracy, which balances how often the model labels things correctly (precision) and how often those labels match expert classifications (recall). Higher scores indicate strong performance; lower scores typically reflect the need for more training data or naturally challenging groups to identify from images.
A full breakdown of accuracy for every group—including simple explanations of precision, recall, and confusion matrices—is available here.
The model can currently identify 17 major benthic groups (the same categories MERMAID uses to calculate benthic cover) and 38 coral genera commonly found on reefs around the world.
In addition, MERMAID AI detects coral growth forms—including branching, foliose, massive, and plating—for key hard coral groups such as Acropora, Montipora, and Porites. These growth forms provide important insight into how corals build reef structure and respond to disturbances, helping users better understand reef health and resilience.
You can view the full list of all labels classified by MERMAID AI here.
No. MERMAID uses a single standardized taxonomy—aligned with the World Register of Marine Species (WoRMS)—to ensure data remain consistent and comparable across sites, years, and projects. All labels used in the MERMAID AI image classification follow this same taxonomy, meaning they use scientifically accepted species and genus names rather than custom codes or user-defined categories.
You can view or download the full reference list in the MERMAID Collect app under Reference in the top toolbar. If you encounter a species that isn’t yet included, you can propose a new addition directly while entering your benthic observation in MERMAID Collect.
MERMAID actively updates its benthic taxonomic reference list to reflect the latest scientific changes. These updates are added continuously, ensuring that new or revised taxa become available for users when entering observations or annotating photos. When users apply these updated labels in their data or annotations, they are stored in the shared training dataset.
As these new or revised taxa accumulate enough high-quality, expert-confirmed annotations, they are incorporated into future versions of the AI model. This allows MERMAID AI model to evolve alongside real-world taxonomic updates and expand the range of groups it can accurately identify.
Once a new model version is released, users can also reprocess older photo quadrat datasets so their classifications align with the updated taxonomy.
MERMAID continuously improves its AI model using uploaded coral reef photos and expert-verified annotations contributed by the community. Before any photo and annotation enters the training dataset, all identifying information—such as project names, usernames, locations, and dates—is removed. This ensures your raw data and project results remain fully protected and visible only within your MERMAID account, under the control of your project admins.
These anonymized photos along with their confirmed annotations are then stored in a public, open-access training data bucket, allowing the global reef science community to benefit from shared data and support transparent, collaborative AI development.
When enough new and diverse photos have been contributed, MERMAID retrains the classifier and releases an updated model to all users. The timing of each new release depends on the volume of contributions—the more photos users upload, the faster the model can advance.
The next model update is expected by early 2026.
You can learn more about the open training data bucket here.
MERMAID uses AI responsibly by prioritizing transparency, collaboration, and sustainability. Rather than having every user train their own model — a process that consumes large amounts of energy and computing power, MERMAID develops shared, open AI models that benefit the entire community. Each model is built from anonymized and verified reef images and annotations contributed through MERMAID, ensuring privacy, fairness, and global representation. We openly share our training data and model documentation to promote accountability and reproducibility, and users always retain full control of their data—no private or sensitive information is ever used for model training. By developing shared generalized models, we reduce duplication, significantly lower the environmental footprint of AI development, and make advanced coral reef image analysis accessible to everyone.