ARTIFICIAL INTELLIGENCE

AI Promises to Revolutionise Quality Control in the Gilthead Sea Bream Industry

Turkey, 4 February 2025 | A groundbreaking study achieving 100% accuracy in quality predictions has heralded a new era for the aquaculture sector

Cabeza de dorada con cebolla

Traditionally, the quality control of fish and seafood has predominantly relied on sensory evaluation methods carried out by individuals. While effective, these techniques are inherently subjective and prone to human error, often exacerbated by the repetitive nature of the assessments which can lead to assessor fatigue. The introduction of methods underpinned by deep learning algorithms and explainable artificial intelligence (XAI) is setting new industry benchmarks, offering solutions that are not only faster and more precise but also distinctly objective.

Researchers from the Isparta University of Applied Sciences and Suleyman Demirel University have spearheaded this innovative approach. Utilising advanced neural networks such as DenseNet121, Inception V3, and ResNet50, they have enabled detailed monitoring of gilthead sea bream under refrigerated conditions by analysing over 3,000 digital images focusing on the eyes and gills.

This state-of-the-art method categorises the freshness of sea bream into three clear stages: fresh, moderate, and spoiled, providing exceptionally accurate predictions. The model's accuracy reaches nearly 100% for the top categories and remains robust at no less than 98.42% when detecting deterioration based on eye characteristics. This approach surpasses the capabilities of traditional methods and offers a non-invasive way to assess quality.

Moreover, the integration of sophisticated analytical tools such as Grad-CAM and LIME has been pivotal in ensuring transparency in the decision-making processes of these AI systems. These tools clarify which features are deemed most crucial by the algorithms, thus enhancing the confidence of quality control professionals and deepening their understanding of how these automated systems formulate their conclusions. This represents a significant advancement towards more reliable and precise industry evaluations.

AI technology is ready to be fully integrated across the entire supply chain, promising transformative changes from aquaculture farms, where real-time monitoring adjusts conditions to ensure optimal growth, to retail points where it guarantees only the freshest fish reaches consumers. High-resolution cameras in processing facilities now quickly assess and categorise fish quality, AI systems in packaging sections automatically classify sea bream based on its freshness, and during the distribution phase, logistical operations are optimised using real-time data.

This technological progress not only boosts efficiency and profitability but also meets vital demands for food safety and environmental sustainability. As this technology becomes the standard in the industry, it is poised to revolutionise seafood quality management, enhancing access to fresh, high-quality seafood, reducing waste, and strengthening consumer trust in marine products.

Related