An autonomous YOLOv9-edge-to-cloud intelligence system for proactive honeybee mortality mitigation

Area of Science:

• Agricultural Science
• Environmental Science
• Computer Science

Background:

• Honeybee mortality is a major threat to apiculture and biodiversity.
• Manual hive inspections are time-consuming and may miss early signs of pesticide exposure.
• Timely detection of pesticide-associated bee deaths is crucial for evidence collection and beekeeper intervention.

Purpose of the Study:

• To develop an automated early-warning system for detecting honeybee mortality.
• To differentiate between naturally senescent bees and those affected by neurotoxic pesticide poisoning.
• To provide rapid alerts for timely on-site investigations and sample collection.

Main Methods:

• Integration of YOLOv9 deep learning with a real-time IoT framework.
• Development of a waterproof beehive with an edge-side image acquisition terminal.
• Training the model on a field dataset and utilizing a lightweight IoT communication protocol.

Main Results:

• The system achieved a mean average precision (mAP@0.5) of 0.937 in classifying honeybee mortality.
• The model detected morphological signs indicative of neurotoxic pesticide exposure.
• The system successfully provided instantaneous mobile alerts via Discord Webhook.

Conclusions:

• The automated system offers a viable solution for early detection of pesticide-induced honeybee mortality.
• This tool complements standard colony management practices by providing timely alerts.
• The system supports rapid response, aiding in the collection of residue samples for analysis.