As total aquaculture production almost tripled its volume during the last 15 years, there is a demand for new and effective methods for cost effective management of the fish farms and disease control. Digitalization and an effective use of big data together with machine learning will make the operators capable of smarter production and increased profits. This project seeks to meet the industry's need for an intelligent, reliable tool that can improve the evaluation and prediction of fish growth and health, and thereby increase revenues and cut operating costs. A study performed at Uni Research Center for Big Data Analysis revealed that a machine learning approach can successfully be used to predict fish growth (weight and length) by intelligent integration of relevant data from different sources. Our tool, named Predict-fit, can, by using an unique and novel machine learning approach, identify the causal and correlation relationship between factors that influence fish growth, health and disease. This suggests that Predict-fit can be used to improve biomass estimation, that will increase operational planning accuracy and profits. This new control of fish health evaluation and prognosis will support proactive measures and reduce loss. Predict-fit is flexible, scalable and can easily be expanded to accept more input parameters, and forecast multiple outputs. Predict-fit seeks 5 MNOK in funding to verify that our novel machine learning method can give accurate biomass estimations and valid fish health predictions for fish farmers.
Project leader: Lars Grønnestad
Institution: VESTLANDETS INNOVASJONSSELSKAP AS