Autonomous Monitoring of Insects (AMI)
Background
The Autonomous Monitoring of Insects (AMI) system is designed to provide real-time, automated monitoring of insect populations, particularly moth species, in Kenya. The project addresses the need for continuous biodiversity assessment in forested ecosystems. Traditional insect monitoring methods rely on manual capture and identification, which are labor-intensive and time-consuming. AMI integrates automatic image capturing, offering a scalable solution for ecological monitoring.
Accomplishments
The AMI system has been successfully deployed at the DeKUT Conservancy, where it has collected high-resolution 4K images of nocturnal insects over a four-month period. The system includes a LepiLED UV light trap, a Logitech BRIO 4K camera, and a Raspberry Pi-based processing unit, all powered by a solar-charged battery system. Initial analysis has focused on insect count estimation, with plans for species identification using deep learning. Additionally, the project has compiled a dataset of moth images from the field and GBIF repositories, setting the foundation for training AI models for insect recognition.
Next Steps
- Retraining a foundation model (SAM – Segment Anything Model) to specialize in identifying moth species.
- Implementing real-time edge processing to reduce data storage requirements and speed up analysis.
- Miniaturizing the AMI system for portability and expanded deployment in diverse ecological zones.
- Integrating advanced error correction techniques to improve data reliability in remote deployments.