A simple smartphone could soon become an essential ally for beekeepers in their battle to control the varroa mite. A mobile application for identifying and counting these tiny parasites in the hive – the first of its kind in Switzerland – has just been developed.
New technologies are supporting beekeepers in their fight against the varroa parasite. Apizoom Sàrl, a Swiss start-up, supported by EPFL and assisted by Agroscope’s Centre for Bee Research in Bern and Identitas AG, has developed a mobile application for automatically counting varroa mites on the varroa tray of a beehive. The application, winner of the canton of Fribourg’s 2020 Agricultural Innovation Award, could become a valuable beekeeping tool, eventually replacing the tedious process of making a visual count of dead varroa mites.
A dangerous, difficult-to-detect parasite
For decades, the beekeeping sector has suffered from a particularly harmful parasite: the Varroa destructor mite. The Swiss Bee Health Service recommends regular monitoring of infestation levels in colonies as a control measure. The reference technique consists in counting the number of dead varroa mites that have fallen onto a varroa tray placed under the hive and separated from the colony by a wire mesh, in order to estimate the overall level of infestation in the hive. However, since the parasite is very small (less than 2 mm), difficult to recognise, at times present in large numbers, and mixed with other waste to boot, this monitoring process is highly labour-intensive. The inspection also requires sharp eyesight, and must be repeated several times during the season.
Artificial intelligence to the rescue of bees
The quality of the photosensors of the latest generation of smartphones opens up new prospects for beekeepers, allowing them to distinguish mites with high precision. The resulting images can be integrated into automatic image analysis processes for automatic counting using artificial intelligence techniques. This is what the Apizoom application offers. The detection model, developed using neural network-type learning methods, is regularly validated and updated.
In the digital age, the mobile collection and centralisation of data with an application like Apizoom enables the systematic statistical analysis of the development of the varroa population in quasi-real time and on a regional or national scale. Accustomed to collective work, our bees would not reject such an approach, especially if the data gathered are used for their benefit.
- Under optimal conditions, the automatic detection model now available approaches human visual acuity.
- The quality of the results depends on the camera used, the light conditions around the apiary and the density of waste on the varroa tray.
- Technological advances and the increasingly large number of images available for training the model will reduce the current limitations and further improve results.
- The centralised collection of data made possible by the application is likely to prove very useful for research and regional monitoring of varroa infestations, leading to the opportunity for collaboration between private stakeholders and government agencies.
- More information about the mobile application can be found on the Apizoom website: https://www.apizoom.app.