Advancing social insect research through the development of an automated yellowjacket nest activity monitoring station using deep learning

Abstract

We describe the development and validation of an autonomous monitoring station that identifies and records the movement of social insects into and out of the colony. The hardware consists of an illuminated channel and a fixed camera to capture the wasps' activities. An ad hoc post-processing software was developed to identify the direction of movement and caste of the recorded individuals. Validation results indicate that the model can detect with high levels of accuracy the presence of workers, drones and gynes, whereas direction of movement is accurate only for workers and drones, but not for gynes. Further development of the software and hardware should enable higher levels of accuracy, especially in terms of the direction of movement of reproductive individuals. This innovative tool holds immense potential for advancing ecological and behavioural research by providing researchers with rapid and easily accessible data. Understanding the activity patterns of individual wasps within the colony can yield valuable insights into factors influencing their growth, foraging patterns and the behaviour of reproductive individuals. Ultimately, this information can be incorporated into effective management plans for controlling harmful social insect populations in both ecological and productive systems.