CoralNet

[ This blog post is adapted with permission from one originally posted as a MERMAID Reef Story ]

Coral reef monitoring is data-intensive work. A single survey can produce thousands of underwater photos, fish counts, bleaching observations, and benthic point estimates. Turning all of that into trustworthy, comparable data has long been one of the hardest parts of reef science.

Over the last decade, two open-source platforms have been developed to tackle different parts of that bottleneck. CoralNet, developed at UC San Diego and the Scripps Institute of Oceanography, brought deep learning to benthic image annotation, freeing researchers from labelling millions of sample points and allowing users to train their own science-ready models. MERMAID, led by the Wildlife Conservation Society, built a field-to-dashboard workflow so reef teams can collect, validate, share, and analyze standardized survey data across fish, benthic, and bleaching methods; with MERMAID AI, a single pre-trained image classification model provides an easy introduction for classifying coral reef images.

What happens when those two platforms meet?

A partnership built on open-source technology

CoralNet and MERMAID share more than a mission to help advance coral reef conservation and science. We share an ethos: that the infrastructure underpinning coral reef science should be open, free, and built for the global community of researchers, managers, and practitioners who do the work.

Over the past two years, CoralNet and MERMAID have been working together to bring our tools into closer alignment. Not by collapsing them into one platform, but by making sure they work seamlessly together. Each is best at what it does:

CoralNet provides broad flexibility for scientific benthic image annotation. Its deep-learning models perform on par with human expert annotators, and users can choose their own label sets and train custom classifiers on their own image libraries.
MERMAID is purpose-built for end-to-end reef monitoring workflows, from offline field collection in remote sites in MERMAID Collect to analysis and data sharing on MERMAID Explore. Trained with CoralNet public imagery and using the same deep-learning feature extractor as CoralNet, MERMAID AI offers an easy-to-use image classification model for users to get started with photo quadrat analysis using a standard set of labels from the MERMAID benthic hierarchy. CoralNet provided the public training data to build MERMAID’s pre-trained AI image classification model. This scientifically validated training data was critical to developing MERMAID’s Image Classification (Beta), released in June 2025.

While MERMAID harnesses the same underlying deep learning methods and feature extractor as CoralNet, the MERMAID classifier is trained with a unified set of labels of the most important taxa, allowing users to directly analyze their images without further training or labelset choices, and to more readily compare results across studies, regions, and time.

For CoralNet users who wish to import their data into MERMAID, we’re pleased that the MERMAID team has developed Easy PQT to make this seamless.

The new Easy Photo Quadrat (Easy PQT) app connects CoralNet data to MERMAID

Easy PQT, or ‘Easy Photo Quadrat’ is a free, web-based app in the MERMAID ecosystem that brings your benthic photo quadrat data from CoralNet sources directly into MERMAID projects — no R, no scripts, no command line. Simply upload your CoralNet export file and start the import process into MERMAID¹.

In a few clicks, you can:

Upload your CoralNet CSV export into the Easy PQT app
Map your labels to MERMAID's standardized benthic attributes and growth forms
Resolve any flagged mismatches with built-in validation
Import your reshaped data straight into a project, ready for final review and submission in MERMAID Collect

Once your CoralNet data is in MERMAID, it sits alongside your fish, bleaching, habitat complexity, and other reef monitoring data in a single standardized format. From there, you can export your data from MERMAID Collect or Explore, analyze it using the MERMAID R package (mermaidr), visualize results through interactive chats in MERMAID Explore, and easily share findings with collaborators or contribute to global reef data efforts.

¹ EasyPQT also supports a ReefCloud connection from the AIMS-led ReefCloud image classification web app.

What this partnership means for monitoring teams using CoralNet and MERMAID

Keep using your own trained CoralNet models. If you've invested time training a classifier on your region, your taxonomy label set, your image conditions, keep using those models in CoralNet! To combine these results with your other MERMAID surveys, export your results and upload them into your broader MERMAID workflow easily with EasyPQT. CoralNet continues to distinguish public and private sources (See Privacy Policy) , and MERMAID has its own data privacy policy that would apply to data imported with Easy PQT.

One place for all your reef data. MERMAID helps you integrate your photo quadrat data with your bleaching, reef fish, benthic (PIT, LIT, or habitat complexity), and soon macroinvertebrate data.

Open, standardized, comparable. Both CoralNet and MERMAID platforms are committed to open data and shared standards. Data flowing through this workflow can support local management decisions and contribute to global reef assessments. CoralNet and MERMAID are currently collaborating on developing a shared ‘open data bucket’ to help improve future image classification models for coral reefs.

Free and open source. Both tools are free for users and built with open source standards, not proprietary software. CoralNet has been supported by NSF, NOAA and the Pacific Blue Foundation; MERMAID by a coalition of foundations and partners committed to conservation tech as public infrastructure.

How to get started

Try Easy PQT: https://datamermaid.shinyapps.io/easyPQT/
Read the documentation on how to use Easy PQT for CoralNet data
Read the documentation on how to use the MERMAID image classification (beta)
Learn more about CoralNet: https://coralnet.ucsd.edu/about/

An open ecosystem for reef science

The reef monitoring community has spent decades building tools, standards, and datasets, often in parallel, sometimes redundantly. The promise of open-source conservation tech is that we don't have to keep working in silos in order to advance coral reef science and conservation.

Acknowledgements

A version of this blog post was written by the MERMAID team, and originally posted as a MERMAID Reef Story (https://datamermaid.org/reef-stories/). We're grateful to Emily Darling, Kim Fisher and the rest of the MERMAID team for a productive collaboration, and to the funders — NSF, NOAA, Pacific Blue Foundation and the philanthropic foundations behind MERMAID — who make this kind of open infrastructure possible. The coral reef community deserves tools that work together, and we'll keep building toward that.