In this paper we presented a novel approach for estimating flower cover, as an indicator of bee abundance and diversity in grassland ecosystems by using UAV RGB images and different machine learning methods. We find highly significant positive relationships between flower cover estimates obtained through UAV RGB images and machine learning algorithms and flower cover estimates obtained the traditional way, in situ by local observers. We also find reliable relationships between UAV RGB image obtained flower estimates and in situ bee abundance and diversity. This represents a proof of concept that imagery from UAVs can be used to reliably assess an important aspect of grassland quality for bees.
jhpoelen
commented
1 year ago
Lots of inferred digital traces based on information about biotic interactions: https://doi.org/10.1016/j.ecolind.2023.110123
From the discussion section: