Where Agriculture and livestock farming meet High Performance Computing
The goal of the Horizon2020 Cybele project is to support precision agriculture and livestock farming by securing access to high performance computing (HPC) facilities and enabling scalable big data analytics. Dr. Steven Davy, of the Telecommunications Software and Systems Group (TSSG), and his team are focusing on the $230bn global aquaculture industry in order to address one pillar of this intention.
Aquaculture is probably the fastest growing food-producing sector and now accounts for more than 50 percent of the world's fish that is used for food. With the world population expected to reach nine billion by 2050, the aquaculture sector will play a key role in ensuring food and nutrition security. However, this growth is not without challenges; in order to satisfy the demand and minimize the impact on the environment, the sector has to use new technologies to intensify, diversify and produce in a more efficient, sustainable and environmentally friendly way. One of the main issues in commercial aquaculture is the lost food when the fish are fed.
Within the Cybele project, aquaculture monitoring and feeding optimisation is one of the nine use cases defined to address the above issues. It aims to ensure that fish food, which accounts for up to 70 per cent of costs on fish farms, is given to stocks at the optimum time and appropriate levels to promote growth and reduce pollution from rotting food debris. It is our intention to make use of drones, image processing and data mining to optimise feeding, evaluate the impact on the environment and evaluate the status of the infrastructure in open sea aquaculture.
This aquaculture use case will make use of the HPC National Supercomputer in TSSG, which is funded by Science Foundation Ireland, to develop computer-vision algorithms to classify images from aerial drones in real-time. This will assist fish farmers in the Mediterranean Sea to assess the health of fish and to optimize feeding by identifying fish behaviour during feeding. The drone cameras and the sensors that are installed in our pilot user sites will be generating large amounts of video and signal data. This data holds information regarding several activities (fish feeding, behaviour, appetite) and incidents (farm status, feed by products diffused to the environment, etc.). Being able to process this data and extract insights and results within short time windows is a big challenge demanding high throughput, computational intensity and short turnaround times. This image processing and analysis is a task highly suited to HPCs and their attributes.
Our ultimate goal under the Horizon 2020 funded CYBELE project is to produce technology that can revolutionise farming, reduce scarcity and increase food supply, significantly boost animal welfare and bring social economic and environmental benefits.