About
FathomNet is a comprehensive AI and machine learning ecosystem designed to accelerate the analysis of underwater visual data. Developed by MBARI (Monterey Bay Aquarium Research Institute), the platform consists of three interconnected tools that together lower the barrier to ocean imagery research. The FathomNet Portal is a globally accessible, end-to-end interface that combines the database and community tools with machine learning capabilities to dramatically speed up ocean imagery analysis. The FathomNet Database is a growing repository of expert-annotated underwater images and pre-trained ML models representing all forms of underwater life, providing machine-learning-ready datasets for both software engineers and marine scientists. FathomVerse is a mobile game that invites ocean enthusiasts, animal lovers, and armchair explorers to help train AI models by interacting with real ocean imagery collected by researchers — making citizen science engaging and impactful. FathomNet is ideal for marine biologists conducting species identification at scale, computer vision engineers building underwater classifiers, and educators introducing students to ocean biodiversity and AI. Its open-access model and community-driven annotation system support reproducible research and collaborative discovery. The platform regularly publishes datasets, participates in Kaggle competitions, and shares resources for educators, making it a hub for the intersection of AI and ocean science.
Key Features
- FathomNet Portal: A globally accessible end-to-end platform that integrates the database and ML capabilities to dramatically accelerate the annotation and analysis of ocean imagery.
- Expert-Annotated Image Database: A growing database of expert-labeled underwater images covering all forms of marine life, providing ML-ready datasets for researchers and engineers.
- Pre-trained Ocean ML Models: A curated library of machine learning models trained on underwater imagery, enabling rapid deployment for species identification and marine data analysis tasks.
- FathomVerse Citizen Science Game: A mobile game that engages the public in training AI models by labeling real ocean imagery collected by researchers, scaling annotation through gamification.
- Community & Educational Resources: Provides annotation guides, educator resources, database help, and a network of marine scientists and programmers collaborating on ocean AI research.
Use Cases
- Marine biologists using the FathomNet Portal to automatically identify and annotate species in hours of underwater video footage.
- Computer vision engineers accessing ML-ready annotated ocean image datasets to train and fine-tune underwater species classifiers.
- Educators integrating FathomNet resources and FathomVerse into marine biology or AI curriculum to engage students with real-world data.
- Research institutions submitting and sharing annotated imagery to expand the global database of underwater life for collaborative science.
- Citizen scientists playing FathomVerse on mobile to help label real ocean footage and contribute to AI model training without technical expertise.
Pros
- Free & Open Access: The platform is freely accessible to researchers, engineers, and enthusiasts worldwide, lowering barriers to ocean AI research.
- Domain-Specific ML Datasets: Offers rare, expert-annotated underwater imagery datasets that are difficult to source elsewhere, making model training for marine applications much faster.
- Multi-Audience Platform: Serves marine scientists, ML engineers, educators, and citizen scientists through distinct but interconnected tools tailored to each audience.
- Backed by MBARI Research: Developed and maintained by a world-renowned ocean research institute, ensuring scientific rigor and high-quality data curation.
Cons
- Narrow Domain Focus: FathomNet is purpose-built for ocean and marine science use cases, making it unsuitable for general-purpose computer vision or non-marine datasets.
- Limited General ML Tooling: Users looking for broad ML experimentation or non-ocean data pipelines will need to supplement FathomNet with additional platforms.
- Dataset Coverage May Vary: As a growing database, coverage of rare or deep-sea species may be limited depending on the current state of expert annotations.
Frequently Asked Questions
What is FathomNet?
FathomNet is an open AI and machine learning platform developed by MBARI that accelerates the analysis of ocean visual data through a database of annotated images, pre-trained models, and community tools.
Who can use FathomNet?
FathomNet is designed for marine scientists, software engineers, educators, and ocean enthusiasts. The platform offers different tools tailored to each audience, from the technical Portal and Database to the citizen-science FathomVerse game.
What is FathomVerse?
FathomVerse is a mobile game that allows the general public to help train AI models by interacting with real underwater imagery collected by researchers, turning citizen science into an engaging experience.
Is FathomNet free to use?
Yes, FathomNet is a free, open-access platform. The database, portal, and associated resources are available at no cost to support global ocean research.
How can I contribute to FathomNet?
You can contribute by submitting localized image annotations to the FathomNet Database, participating in the FathomVerse mobile game, joining the mailing list, or engaging with the community on social media.
