Data science is applied to extract knowledge and insights from various kinds of data, and apply knowledge and actionable insights from data across a broad range of application domains [1]. As computing power is becoming much stronger, data science can be applied to solve more real-life problems with a large amount of data, e.g. smart city, autonomous driving, recommendation system, safety and security system. In the EU-funded research project CYBELE, we focus on some specific use cases in agriculture by developing efficient and more intelligent applications to improve production.
Fish is one of the most popular food we have every day. HLRS and ILVO are together working on training a neural network with collected data during fishing such as path and other corresponding geographic information to predict the weight of different fishes to be caught, as Figure 1 shows. It could also help to find the optimal path in some specific fishing areas in order to get the target weight of fish in the subsequent fishing. In other words, fisheries are able to better plan fishing activities depending on demands. Moreover, the price of fish could also become more favourable when more cost and time could be saved while fishing.
One of the biggest challenges for this project is data collection. Collecting data while fishing is a time-consuming process and therefore very costly. At the moment, we are testing different methods to augment the data, and are performing simulation of fishing process in HLRS HPC systems.
Fig. 1: Historical fishing path (blue) and corresponding weights are collected to train a neural network which predicts fish weight depending on the current fishing path (red).
[1] https://en.wikipedia.org/wiki/Data_science