Coral Reef Health Monitoring

Detecting and identifying coral species in benthic imagery to track reef health over time.

Manual

Image Selection

Choose an image from the examples provided below, or upload your own data.
Run the ML model

Click the 'Submit' button to initiate the machine learning model.
Visualize the results

The system will generate bounding boxes and segmentation masks for the detected coral species, each accompanied by a corresponding probability score.

Overview

Coral Reef Health Monitoring brings computer vision to benthic (seabed) imagery, detecting and segmenting corals so reef surveys can be analyzed in a fraction of the time. Built in collaboration with ReefSupport, it turns the photos and video collected on research dives into measurable data on what lives on the reef — and how that changes over time.

Marine biologists spend a large share of their time manually processing dive footage. By automating the segmentation step, this tool removes the reporting bottleneck and helps quantify the long-term growth or decline of coral cover within marine protected areas.

How It Works

From a single benthic image to quantified reef cover:

Benthic image input

Works with the imagery already gathered on research dives — pick an example or upload your own seabed photo.

Detection & segmentation

Draws bounding boxes and pixel-level segmentation masks around each coral, isolating it from sand, water, and other substrate.

Probability scores

Every detection comes with a confidence score, so analysts can prioritize review and keep only reliable results.

Coral-cover quantification

Segmentation masks translate into measurable coral cover — the foundation for tracking reef change survey after survey.

Why Reef Monitoring Matters

Biodiversity hotspots

Reefs shelter roughly a quarter of all marine species on under 1% of the ocean floor — monitoring their health protects an outsized share of ocean life.