Queen detection
| goal / problem solved | 💀Colony without a queen |
|---|---|
| Status | Groomed |
| billing plan | 1-community |
| verticality | vertical |
| concept hierarchy path | apiary/hive |
| Date | |
| Priority | 14 |
| Tasks | Change queen detection to need a confirmation from a user |
| estimate | 5 |
| impact = priority % | 70 |
| Tags | MLbackendfrontend |
Background
Queenless colony in a good scenario will create queen cups and create new queens. In worst case it can lead to workers laying eggs. So if beekeeper would get alerted, he should intervene and add a new queen.
Goal
We want to detect queen on beehive frame to know if bee colony is queenless or not, as this can kill a colony if left untreated.
We currently use https://github.com/Gratheon/models-bee-detector model that can detect queens, but its precision is quite poor. It is based on https://universe.roboflow.com/matt-nudi/honey-bee-detection-model-zgjnb dataset and weights. We need a good precision to not get false-positive results.
AC
- have high-precision bee queen detection AI model
- integrate it in web-app