New grader with precise and efficient grading of flatfish
Aquaculture is one of the world’s fastest-growing food-production sectors, expanding at an impressive annual rate of 5.3 % over the past decade [1]. Within this arena, farming flatfish such as sole (Solea solea) and turbot (Scophthalmus maximus) commands special attention because of their high market value and strong demand in gourmet cuisine. Yet producers face significant hurdles, especially during the juvenile phase, where precise size grading and weight estimation are critical for uniform growth and maximum survival.
The challenges of grading
Current hatcheries rely mainly on semi-automatic graders that, although superior to manual methods, still show major limitations. Most depend on a single parameter —either weight or height — thereby restricting sensitivity and accuracy. Systems based solely on weight can miss key morphological traits such as length or width, while height-based systems overlook body volume and condition.
This single-criterion approach can yield inconsistent results: two fish of identical weight may have very different shapes, influencing later growth and tank performance. Moreover, these machines are invasive, requiring physical handling that raises stress levels and can increase mortality by up to 15 % in some cases [4].
The promise of artificial intelligence
These shortcomings underscore the need for advanced solutions capable of integrating multiple morphometrics in real time. Artificial intelligence (AI) provides that path [2]. Deep-learning techniques—particularly convolutional neural networks (CNNs)—have proven highly effective for image classification and segmentation [5], and are already being used to detect fish diseases and estimate biomass [6].
For juvenile grading, AI can combine simultaneous measurements of length, width and projected area to infer weight without physical contact, delivering real-time data that support adaptive production management [7].
Benefits beyond efficiency
Intelligent grading improves more than throughput: it enhances both environmental and economic sustainability. Precise size segregation optimizes feed allocation [3], reducing one of aquaculture’s largest cost centers while limiting waste. Continuous, high-resolution data on juvenile cohorts empower operators to fine-tune feeding regimes and schedule transfers to grow-out facilities at optimal times [8].
FishFarmFeeder: A step toward the future
A concrete illustration is the intelligent grading platform from FishFarmFeeder, which merges machine vision with machine-learning algorithms for real-time, size-based sorting of flatfish.
The system aims to:
- boost productivity,
- cut operational costs and
- furnish statistical insights for refined husbandry and feeding strategies.
The architecture employs distributed processing:
- Sampling and parameterization – statistical analysis of a representative fish sample establishes grading thresholds.
- Continuous operation – every individual is graded and its weight estimated as it traverses the line, with results available instantaneously for decision-support dashboards.
Conclusion
AI-driven automation is not merely a promising fix for today’s bottlenecks; it is a stepping-stone toward greater sustainability and profitability in aquaculture. Systems such as Fish Farm Feeder’s demonstrate how technology can transform food production, fostering a more efficient and resilient future for the sector.
References:
[1] FAO. (2022). The State of World Fisheries and Aquaculture 2022. Food and Agriculture Organization of the United Nations.
[2] Yang, Xinting & Song, Zhang & Liu, Jintao & Gao, Qinfeng & Dong, Shuanglin & Zhou, Chao. (2020). Deep learning for smart fish farming: applications, opportunities and challenges.
[3] Papandroulakis, Nikos & Dimitris, Papaioannou & Pascal, Divanach. (2002). An automated feeding system for intensive hatcheries. Aquacultural Engineering. 26. 13-26.
[4] Føre, M., Frank, K., Norton, T., Svendsen, E., Alfredsen, J. A., Dempster, T., … & Berckmans, D. (2017). Precision fish farming: A new framework to improve production in aquaculture. Biosystems Engineering, 173, 176-193.
[5] Hasan, N. (2022). Fish diseases detection using convolutional neural network (CNN). International Journal of Nonlinear Analysis and Applications.
[6] Haddad, Dr & Mohammed, Fatima. (2024). A Convolutional Neural Network Approach for Precision Fish Disease Detection. EVOLUTIONARY STUDIES IN IMAGINATIVE CULTURE. 1018-1033.
[7] Kandimalla Vishnu , Richard Matt , Smith Frank , Quirion Jean , Torgo Luis , Whidden Chris. (2022). Automated Detection, Classification and Counting of Fish in Fish Passages With Deep Learning. Frontiers in Marine Science, 8.
[8] Naylor, R. L., Hardy, R. W., Buschmann, A. H., et al. (2021). A 20-year retrospective review of global aquaculture. Nature, 591(7848), 551-563.
Javier Álvarez Osuna is a director of R&D and information technologies at FFF.
His 30 years of professional activity have been developed mainly in the technological field and specifically in R&D. He has been a Torres Quevedo researcher (2009) and his merits include participation in more than 25 R&D projects and management of another 10 as principal investigator.
He is the author of 10 international articles and holder of 2 international technological patents. He is an expert in the development of embedded systems and software for industrial environments that require high performance.
Doctor in Pharmacy – the University of Santiago de Compostela (1997).
