Huge fields and new satellite images every couple of days—that means vast volumes of data. Farmers need systems that enable them to analyze this information at high speed. And that was the aim of the EU project DataBio, which concluded in December 2019 after three years’ work.
The goal was to leverage artificial intelligence to improve yields and sustainability in agriculture, fishing, and forestry. At Fraunhofer IGD in Darmstadt, the focus was on farming. “All three elements of the project combine multiple data sources,” explains Ivo Senner from the Spatial Information Management Competence Center. In this way, it is possible to identify and make use of correlations.
“Particularly with large expanses of land, it can be difficult to identify invasive species or environmental degradation,” the IT expert states. “That makes it complicated, for instance, to assess loss of crops in order to calculate compensation.” Ideally, the new systems will enable the effective evaluation of even very small areas. Moreover, it is hoped they will support EU-common agricultural policy processes designed to verify the veracity of applications for funding. The new technology will, for example, allow plausibility checks across significant territories instead of making spot checks, as is currently the practice. This is made possible by incorporating efficient processes drawn from artificial neural networks. “Artificial neural networks are computer systems that mimic the structure of the human brain. And just like human brains, these systems can learn. The more often they receive certain signals, the more precisely they are able to recognize images or patterns.” Every one to two weeks, satellites provide new measurements and images—huge amounts of data that can be rapidly processed by artificial intelligence.
The DataBio project entails the participation of 48 partners from 17 countries, including researchers, farmers, and IT companies. Some of the findings are already being employed in practice; others are currently being studied further with regard to their real-world applicability, including, e.g., within the scope of a pilot project in Greece. Talks are being held on possible follow-up projects that will take a closer look at forestry. The methods already developed could then be leveraged to monitor the health of woodlands.