Automated video monitoring of insect pollinators in the field - Royal Botanic Gardens, Kew research repository
Skip to main content
Shared Research Repository
Journal article

Automated video monitoring of insect pollinators in the field

12 February 2020

Abstract

Ecosystems are at increasing risk from the global pollination crisis. Gaining better knowledge about pollinators and their interactions with plants is an urgent need. However, conventional methods of manually recording pollinator activity in the field can be time- and cost-consuming in terms of labour. Field-deployable video recording systems have become more common in ecological studies as they enable the capture of plant-insect interactions in fine detail. Standard video recording can be effective, although there are issues with hardware reliability under field-conditions (e.g. weatherproofing), and reviewing raw video manually is a time-consuming task. Automated video monitoring systems based on motion detection partly overcome these issues by only recording when activity occurs hence reducing the time needed to review footage during post-processing. Another advantage of these systems is that the hardware has relatively low power requirements. A few systems have been tested in the field which permit the collection of large datasets. Compared with other systems, automated monitoring allows vast increases in sampling at broad spatiotemporal scales. Some tools such as post-recording computer vision software and data-import scripts exist, further reducing users’ time spent processing and analysing the data. Integrated computer vision and automated species recognition using machine learning models have great potential to further the study of pollinators in the field. Together, it is predicted that future advances in technology-based field monitoring methods will contribute significantly to understanding the causes underpinning pollinator declines and, hence, developing effective solutions for dealing with this global challenge.

Files

This is a metadata only record.