Automated pollinator identification, community and interaction monitoring from video captures


In order to understand the processes behind the pollinator declines, besides monitoring species occurrences, we also have to focus on monitoring interactions. In addition to traditional trapping and observation methods, we should develop fast, reliable, and cheap automated methods which are capable of monitoring both species occurrences and interactions. Using cutting-edge tools in ecology, such as video analysis with deep-learning-based object detection can be the way forward. Indeed, besides providing information at the community level, video surveillance can also automatically record species interactions in real time with an easily standardisable sampling effort.

We believe, this study can help to make artificial intelligence-facilitated monitoring a trusted tool in pollinator studies.