New case study: SEA-AI - using AI to improve seabed surveys

Improving seabed monitoring through standardised imaging and AI analysis

Regulators such as SEPA and NatureScot must ensure that new and expanded fish farms do not harm sensitive seabed habitats known as Priority Marine Features (PMFs). Undertaking these assessments means relying heavily on underwater video surveys and seabed imagery. Recent regulatory changes have increased the amount of seabed footage required for farm applications - up to three kilometres for expansions and even more for new sites.

Reviewing the video footage manually is time-consuming and costly, and poor-quality video can mean surveys need to be repeated. This creates delays for both industry and regulators, highlighting the need for more efficient, standardised approaches.

The SEA-AI project aimed to address this issue. Valued at £482k, the partners were the University of Highlands & Islands, Scottish Sea Farms, SEPA, NatureScot, Bakkafrost Scotland, and The Data Lab. The project was support by SAIC. Its goals were to:

1. Improve image capture by developing a standard operating procedure that ensures consistent, high-quality footage suitable for both human and automated analysis.
2. Develop AI-assisted tools to identify key biological features linked to PMFs quickly and accurately, reducing the burden of manual review.

The team tested how factors like camera configuration, lighting and survey speed affect image quality, using a remote operated vehicle (ROV) fitted with cameras and lights. The project advanced the technology readiness of machine-assisted seabed monitoring and demonstrated that AI can detect certain PMF features, producing georeferenced maps from typical survey footage.

While human oversight remains essential, SEA-AI could offer a scalable, repeatable approach that improves efficiency and reduces costs. It also sets the stage for future developments, such as broader species detection and integration with autonomous survey platforms. As underwater imaging becomes more accessible, SEA-AI represents a major step toward technology-enabled environmental monitoring that benefits both regulators and industry.

The full title of this project is ‘Improving seabed monitoring through standardised imaging and AI analysis: SEA-AI’.

Read the full case study