BeeScanning on AI-map
BeeScanning has qualified as one of Swedens quality AI-startups. https://aistartuplandscape.se [...]
FREE DETECTION
OF VARROA MITES
Monitor your bees health and detect parasites.
All in your camera.
Know your bees
Find varroa, wingless bees, the queen and colony strength. Accurate, fast and easy to use.
Use your camera! Take photos of your bees on the comb. Analysing is done in 100 milliseconds by AI*.
Get result to determine need of treatment. Use stats to select for breeding, documenting and management.
*Artificial intelligence in neural networks, NN, with deep learning technology.
★★★★★
“Version 1.0.7, magically fast and safe diagnosis of varroa, dwv and queen”
– Google Play User
★★★★★
“Great app. Easy to use.”
– Google Play User
★★★★★
“Fast and easy to use”
– App Store User
Your vote! Next in app v 3.0 ? (read more in FAQ)
How to use
A five minutes tutorial on how BeeScanning is used in real-time, showing its lightness and easy-to-use functions.
We need bees, bees need us
20 million bee colonies (out of 100 millions) dies every year, that’s horrendous! Parasite varroa destructor is a major cause. BeeScanning started in 2016. We found a correlation between what’s visible in an image and how many varoa mites you get from alcohol sampling.
How it works
BeeScanning is an app that enables beekeepers to instantly diagnose varroa infestation. Images taken of live bees on the broodframe are automatically analysed with artificial intelligence. Result is calculated in %, number of varroa/number of bees. A factor is used to multiply the result compensating for hidden varroa. Factor is derived from comparing results with alcohol washing.
It is recommended to take images from distance 21 cm from 2-3 broodframes, covering 3 000 – 4 000 bees. The lower infestation the more bees/images are needed. Infestation below 0,5% may be detected. A way to measure the accuracy of the technology is the recall-parameter, calculating ratio between true positive varroa and false negative varroa. It is close to 80% meaning it finds 8 out of 10 visible varroa. Models are continuously improving. 80% is considered good enough for practical purposes, warning the user in time.
Differences in respect to other systems
BeeScanning is a unique technology based on our database, worlds largest, of manually annotated images of bees. Cutting edge, proprietary software, i.e.
Neural Networks and Deep Learning is used, as opposed to image library technologies. Users images and results are revised and sent back to the system for ever increasing improved performance. The technology offers a unique option by collecting metadata as colony strength, breed, gps, time, infestation, fertile queen. Combining these and future features as other health indicators, brood diseases, nutritional status will bring tools for prognoses and new understanding breeding healthy bees.
Besides finding varroa, deformed-wing-virus and the queen, project is developing analysing another 13 classes to be possible to detect, american foul brood for instance.
In 2020 focus is on analysing video sequences and is already feasible as can be found here. Video has the potential to add to the understanding of how the parasite-host system works as well as reveal new ways to analyse indicators of infestation.
Examples of success
BeeScanning is a fast, nondestructive method as opposed to conventional methods where one kills bees, avoiding the Queen or analyses dead fallen varroa in recurring intervalls. The technology is patent pending and has received 15 international recognized awards. BeeScanning is used worldwide both as a practical tool for the beekeepers and in research where it enables both analysing and documenting.
Björn Lagerman, Founder of BeeScanning
*Use the BeeScanning app for free and contribute to the future of beekeeping. Your images will add to the knowledge database.
