The artificial intelligence that detects the varroa mite

The artificial intelligence comes to the aid of the bees by finding the varroa mite, their number-one enemy, which attacks them in the hives and kills them. Some students of the Federal Polytechnic of Lausanne (Epfl), led by the engineer Jean-Philippe Thiran, have developed a system that identifies the parasites as soon as they attack the hives, with the aim of intervening in time with the treatments useful to defeat them.



Up to now, farmers have been looking for mites on a wooden board under the hive to identify the first signs of infestation. However, it is not an effective technique because the parasites measure just one millimetre and, if the beekeeper has many hives, it takes a long time. The students, encouraged by the beekeeper Alain Bugnon, then created an App with which the beekeeper can upload the image on a web platform. The image is then analysed by the artificial intelligence that identifies the parasites in a few seconds.

In order to allow the system to recognize mites quickly, the students trained it on many images. In particular, they created a database of photos of Varroa mites, to allow the system to recognize pests without making mistakes. Then the system was trained to analyse the photos sent by beekeepers.