BeeMonitor: Automated IoT video surveillance and AI-powered analysis of solitary bee nesting behavior

Pollinators, including bees, play a crucial role in plant reproduction and food production, yet populations of many species are declining due to habitat loss, pesticide exposure, and climate variability. Monitoring pollinator nesting and foraging behavior is essential for understanding their responses to environmental changes, but manual observation is labor-intensive and error prone. In this study, we introduce BeeMonitor, an automated video surveillance and AI-powered processing system for monitoring solitary bees that nest in above-ground cavities. The system integrates hardware for continuous video recording and a software pipeline utilizing YOLOv8 object detection and BeeTrack, a custom multiple-object tracking algorithm tailored for tracking fast-moving bees. We evaluated BeeMonitor by recording and analyzing 110 minutes of monitoring footage and comparing its performance against manual observations. Our results show that BeeMonitor achieved a high recall of 92.9% and a precision of 80.6%. The BeeMonitor system offers a scalable and efficient approach for researchers and conservationists to monitor solitary bee activity, contributing to ecological studies and pollinator conservation efforts.