Data Science for Feed Optimisation for Aquaculture

Every aquaculture farmer knows that feeding and health observation of fish stock is the greatest cost of production in aquaculture. Ensuring each and every fish is optimally fed, healthy and environmentally safe is a massive cost to farmers. Rearing healthy fish requires a delicate balance between overfeeding at the cost of feed wastage, or underfeeding at the cost of sub-optimal fish growth. Our powerful machine learning algorithms extract real-time analytics of the behavior and movement of aquaculture stock. Our approach combines video streams with existing sensory data platforms such as dissolved oxygen, pH, salinity, temperature and other sensors to calculate optimal feeding portions. With this information, farmers can make data-driven decisions related to on-demand feeding, the health of the stock, and the environmental impact of the farm. We augment our machine learning algorithms using computer vision to quantify the feeding behavior, movement, and the growth rate of fish stocks. Combining these data with readily available environmental measurements, our system can provide real-time informed recommendations on optimal feeding strategies, as well as flagging health issues within the stock.