Getting Started

Discover Image Classification (Beta) in MERMAID Collect, our first feature powered by MERMAID AI.

Through a partnership with CoralNet from the University of California San Diego and Scripps Institution of Oceanography, MERMAID now includes AI-based support for benthic image analysis. Our Image Classification (Beta) feature leverages this collaboration, allowing you to streamline and standardize the processing of benthic photo quadrats. Upload entire transects, apply automatic point-based classification using a pre-trained model, and interactively review results—all within an efficient workflow from your project in MERMAID Collect.

Each photo is classified using a fixed grid of 25 points and a standardized set of benthic attribute labels, supporting data comparability across datasets and projects within MERMAID.

AI Image Classification interface showing a photo quadrat with 25 overlaid points and benthic attribute predictions
AI-based Image Classification (Beta) in MERMAID Collect.
Important: Beta Feature and User Review

  • Image Classification (Beta) is undergoing testing and user feedback integration. We encourage you to report performance issues or classification inaccuracies to support ongoing model improvement by contacting us at contact@datamermaid.org.

  • While the MERMAID AI model provides an efficient initial classification, all point classifications must be reviewed and confirmed by a trained observer. This includes verifying model-classified labels and manually classifying any unclassified points to ensure data accuracy and reliability.

High-Level Workflow

The process is designed to be straightforward:

  1. Enable Image Classification (Beta) for your benthic photo quadrat sample unit.

  2. Upload your quadrat photos.

  3. MERMAID AI automatically classifies 25 points in each photo.

  4. Review and confirm all 25 points for each photo, making adjustments as needed.

  5. Submit your validated data.

Flowchart depicting the process of AI-image classification in MERMAID from creating a new sample unit to submitting the data
AI-based Image Classification (Beta) workflow in MERMAID Collect.
Ready to try it out? Here’s how to quickly enable and use the feature:
  1. Navigate to your project in MERMAID Collect and go to the Collecting page.

  2. Click +Add Sample Unit and select Benthic Photo Quadrat.

  3. Click on Classify photos with AI (Beta).

  4. Complete the required fields in the Sample Event and Transect sections. The “Number of quadrat points” will be automatically set to 25.

  5. Click Upload Photos and add your images. The MERMAID AI model will begin processing immediately.

Screenshot illustrating where to enable AI image classification for a new benthic photo quadrat sample unit
Try the new Image Classification (Beta) by enabling the feature when adding a new benthic photo quadrat sample unit.

For more detailed steps, please see sections below in “Using Image Classification (Beta) for Benthic Photo Quadrats”.