Students from Switzerland have trained the ApiZoom image recognition algorithm to search for images of dangerous ticks that parasitize bees. Now beekeepers in search of traces of dangerous and small parasites have to inspect the waste of the hive daily.
One of the new threats to honey bees is the parasite mite Varroa destructor. It lays eggs in the hive and feeds on the hemolymph of bees and their larvae. Although the bees do not die from this, they become weak and sick, and over time the whole swarm dies. Worst of all, it is extremely difficult to notice the presence of ticks – they are no larger than 1.8 mm in size. Therefore, infection often goes unnoticed until it is too late.
Beekeepers are trying with all their might to avoid ticks. To do this, they have to inspect daily the waste of the hive in search of tiny parasites. This is a difficult and time-consuming job.
The way out can be machine learning. If this technology copes with any task perfectly well, it is with sorting noisy data – just such as the surface covered with tiny objects, among which there may be corpses of ticks, TechCrunch writes.
The project of Swiss students started in 2017, and since then the model has been trained in tens of thousands of images.
The percentage of successful attempts to detect parasites reached during this time 90%, which is comparable with the level of human attention. But AI works much faster.
Now students of the Federal Polytechnic School of Lausanne are planning to distribute their application as widely as possible among beekeepers. In the near future, they will offer two solutions – for websites and for smartphones. With the help of applications, it will be possible not only to find out if this hive is infected, but also about the situation with mites in the region. By collecting such data, researchers will be able to track potential Varroa destructor mutations.
An extract from tinder mushrooms of the genus Fomes and Ganoderma will help to fight the Varroa tick. They reduce the concentration of viruses that carry these mites dozens of times.