See like a bee
Bees are used as vectors for pollination and agricultural chemical transport in farmland and greenhouse systems. However, there are some situations where natural pollinator use is problematic, such as hybrid production, pharmaceutical drug cultivation, chemical transport above the tolerance of the vector specie, or atmospheres low in dioxygen.
Small unmanned aerial vehicles (UAVs) could act as an alternative vector in such situations while performing other functions (i.e. yield estimation, early pest detection, soil moisture estimation, etc.). One step towards such a tool is to equip a UAV with flower classification and analysis ability. Traditional drone cameras use a Bayer filter covering the sensor to create a color image. However, this setup is optimized for human vision and not to detect floral cues needed for pollinator vision.
We aim at creating a digital vision system optimized for the UV range of a bee’s visual spectrum; which can be incorporated into a drone. It will demonstrate the ability of such a system to distinguish crops from weed species, locate the desired plant species, count and map flowers, and identify flowers to the cultivar level.