Species-selective edge-AI insect identification for consumer traps
On-device, species-selective insect identification engine for consumer traps and smart monitoring devices. Uses field-validated computer vision to distinguish target pests such as Colorado beetle from beneficial insects, running entirely on an integrated edge-AI camera. Enables low-cost, chemical-free pest monitoring with species-level activity trends and alerts via Field OS dashboard.
Overview
Computer-vision engine that identifies insect pests to species level on a single integrated edge-AI camera, with no cloud dependency and no professional installation. Built for consumer traps and low-cost monitoring devices in gardens, lawns, and greenhouses. The system is already field-validated for Colorado beetle in cluttered natural backgrounds and variable lighting. Its key differentiator is species selectivity: target pests are flagged while pollinators and other beneficial insects are ignored, reducing false alarms and enabling chemical-free pest management. Detections can optionally flow into the Field OS dashboard for species-level activity trends and alerts.
Technical specifications
The in-house species-classification model is being ported to the CamThink NeoEyes NE301 integrated edge-AI camera. Work includes:
- Benchmarking accuracy, latency, and power on consumer-grade hardware
- Optimizing duty cycling for long unattended operation
- Fine-tuning for additional target species
- Controlled outdoor and greenhouse trials measuring precision, recall, false-alarm rate, and detection time
- Data integration via MQTT with Field OS and consumer apps
The on-device architecture keeps identification local, preserving privacy and reducing connectivity requirements.
Technology readiness level
The recognition engine is field-validated for a target pest and is now being adapted to a camera-integrated consumer device. Planned validation: months 0–4 hardware integration and additional target species; months 4–9 controlled trials; months 9–12 integration and licensing preparation. Commercial path is targeted within 3–5 years, with the validated model and software ready for licensing or supply.
About GuardenAI
GuardenAI is a technology startup based in Budapest, Hungary, focused on developing autonomous robotic solutions for the agricultural sector. Founded in 2023 by Nandor Kelecsenyi and Akos Csik, the company specializes in artificial intelligence-powered robots designed to identify and eliminate agricultural pests, such as those causing potato blight, using laser technology. By integrating advanced computer vision with precision robotics, the platform aims to provide a sustainable alternative to chemical pesticides, aligning with environmental goals to reduce the use of harmful agricultural substances while alleviating the physical workload for farmers.
This technology is particularly significant for farmers and the broader agricultural industry as it offers a cost-effective, labour-saving tool for precision crop protection. The innovation has been recognized for its potential to support more sustainable food production, including winning the inaugural CHARM-DEN award for its application in combating plant diseases without chemicals. As an early-stage company supported by the Startit@K&H incubator, GuardenAI continues to develop its systems to address modern challenges in crop health and regulatory compliance, providing a technological bridge toward more environmentally friendly farming practices.
Inquire about this partnering listing
Looking for
Industry applications
delivered to your inbox.