Monitoring fish health in land-based fish farms using artificial intelligence

Swedish fish production needs to increase to meet demand and reduce dependence on imports. At the same time, there are growing requirements for animal health and welfare in aquaculture. This project is developing new AI-based methods to monitor fish health in land-based farms, providing farmers with better tools to detect problems early and ensure sustainable production.

Demand for sustainably produced fish is increasing, but today the majority of the fish consumed in Sweden is imported. Land-based recirculating aquaculture systems (RAS) offer a resource-efficient way to produce fish domestically. For these farms to become both competitive and sustainable in the long term, new methods are required to safeguard fish health and welfare.

The project is developing three different AI-based systems for continuous and non-invasive monitoring:

• Automatic classification of Observational Welfare Indicators (OWI) through underwater imaging.
• Analysis of movement patterns to link fish behaviour to stress and environmental changes.
• Data integration and correlation, where physiological parameters (e.g. heart rate and cortisol levels), water quality, and behavioural data are combined to provide a comprehensive picture of fish health.

The system is being developed and evaluated both in a research environment at the University of Gothenburg and subsequently in a production environment. By combining sensors, AI, and expert assessments, fish farmers may in the future receive early warning signals of ill health, reduce disease outbreaks, and improve production outcomes.

The project contributes to both increased food security and animal welfare and aligns with the EU’s goals for sustainable aquaculture. It also strengthens Sweden’s ability to become more self-sufficient in fish while offering consumers a product with a strong sustainability profile.