Tracking bee decease with AI

In the OCELI project, researchers want to use AI to collect reliable data on the causes of the decline in bee and bumblebee populations. As part of the project, five partners led by the FZI Research Center for Information Technology, are developing technologies to protect bees and bumblebees.

There are many reasons why bee and bumblebee populations are declining in Germany and around the world. In addition to the use of pesticides, other reasons include monocultures, pathogens and destroyed habitats. The species extinction crisis is just as important as the climate crisis, said German Environment Minister Steffi Lemke recently.

One fact is certain: Insect extinction is a major problem of our time. A consortium from science and industry now wants to create a basis for answers to the question »How can the dying be stopped?« in the »OCELI« research project. There have been national and international efforts in the past to halt the decline in biodiversity. Although many of the causes of species loss are known, knowledge about the complex interaction of individual factors is incomplete.

For this reason, the project partners apic.ai, Eurofins Agroscience Services Ecotox, Disy Informationssysteme GmbH and the Helmholtz Centre for Environmental Research – UFZ, together with the FZI Research Centre for Information Technology, now want to create the technical basis to close existing knowledge gaps. Together, they want to research which stressors exist for bee and bumblebee colonies and how they are connected. Among other things, competencies from applied artificial intelligence and sensor technology will be used. They will enable the consortium to precisely measure the effects of the design of agriculturally used areas on pollinators for the first time and to systematically collect data over long periods of time.

Christoph Zimmermann, department head at the FZI, explains: »We want to track in real time how protective measures for pollinators are affecting them. We get support from the honeybees and bumblebees themselves. Their feedback enables us to generate a robust database for the first time, so that targeted measures for insect-friendly agriculture can be developed.«

Data should close knowledge gaps

To collect the data, networked camera systems will be installed at the entrance of beehives or bumblebee colonies this spring, continuously filming all incoming and outgoing animals. Here, apic.ai brings its expertise in visual, local monitoring technology for pollinators. They use neural networks to record and process the recorded activities. Here, FZI is responsible for developing the feature extraction algorithms.

The evaluation takes place at the UFZ by means of the simulation models »BEEHAVE« and »Bumble-BEEHAVE« and should provide information about which dangers and interactions exist. In addition, the analysis of geo-, weather-, land use- and flight monitoring data will establish causal relationships between changes in the environment of the colonies and their development. In doing so, Disy analyzes observational monitoring results. Accompanying field studies, carried out by Eurofins as part of the project, can be used to test hypotheses about specific causes of insect death. Based on the data, it should be possible to develop best practices and effective measures for pollinator-friendly agriculture in the future.

For example, the technology developed in the research project can be used to find out which types of plant protection products are critical or non-critical for pollinators and whether the abundance and diversity of local blooming plants allow pollinators to live well. In addition, the success of specific measures to protect pollinators will be measurable. More information about the research project can be found on the project's website.