BeeVS
Artificial intelligence helps to select Varroa tolerance trait

During autumn/summer 2018, the Austrian beekeeping federation Biene Österreich tested two recently developed systems for the automatic assessment of Varroa infestation in beehives. Precision and accuracy of the systems were measured and compared with the performance of skilled human operators. BeeVS, a portable scanner and Varroa Counter, a smartphone application, were tested according to the Guidelines of the European Medicines Agency (EMEA) for bioanalytical method validation and documented according to Good Field Practice guidelines by the OECD.

The smartphone application, run on devices with cameras of different quality (5MPx und 20 MPx resolution), showed no significant correlation between the true number of mites on sticky boards and the measured value. The application was not able to assess the true number of mites on sticky boards exposed for 5.2 days in the beehive.

In contrast, BeeVS tested under the same conditions, showed a very good correlation between the true number of mites on sticky boards and the measured value. The precision of this device was far superior to that of human operators and it performed slightly better in terms of accuracy. The average amount of time saved amounted to 2.7 minutes per sticky board. The device thus provides results that are at least as correct as those of human operators but does not suffer from the severe lack of repeatability, as is the case when sticky boards are counted by human operators.

In view of the planned digitalisation of the performance testing program by the Austrian breeding program, the employment of this tool would improve the assessment of the Varroa tolerance trait. [/av_textblock] [/av_one_full]