SIP3
Drone-based AI detection of Colorado Potato Beetle
for targeted pest control
Sector
Arable / Potatoes
Country
Belgium
SIP Leader
EV ILVO
Farmer partner
ISP
The challenge
Potato farmers in Belgium apply pesticides to their fields twice every season as a standard practice, regardless of whether the Colorado Potato Beetle is actually present. This blanket approach increases costs, accelerates pesticide resistance, and risks contaminating ground and surface water.
SIP3 set out to replace this routine with precision: autonomous drone scouting that detects exactly where the beetle is, at what growth stage, and in what density, so treatment is targeted only where it is needed.
How the system works
An autonomous UAV flies a pre-planned route over the potato field, capturing high-resolution images of the crop canopy. As it flies, an AI detection model analyses the imagery to identify Colorado Potato Beetle and larvae at different growth stages, locating infestations with spatial precision.
The system generates infestation maps showing exactly which parts of the field are affected and how severely. These maps are then converted into prescription maps that guide targeted pesticide application, treating only the zones where the pest is confirmed present.
The system was tested in two modes: a Cloud mode where images are uploaded to a remote server for processing; and an Edge mode where detection happens directly on the drone during the flight, with only compact results transmitted.
UAV autonomously surveys the potato field, capturing high-resolution crop imagery.
AI mode detects Colorado Potato Beetle at different life stages, mapping infestation zones.
Infestation map generated. Prescription map created for targeted spraying.
Farmer applies pesticide only in affected zones. Unaffected areas left untreated.
Two deployment modes tested
CLOUD MODE
Images are uploaded from the drone to a remote server for processing. Delivers slightly higher detection accuracy and enables centralised data management. Requires stable uplink connectivity during or after the flight.Images are uploaded from the drone to a remote server for processing. Delivers slightly higher detection accuracy and enables centralized data management. Requires stable uplink connectivity during or after the flight.
Best for: areas with reliable connectivity, farms where centralised data management is a priority
EDGE MODE
Detection happens on the drone during the flight using an onboard AI accelerator. Only compact results are transmitted. Works in fields with no connectivity.
Best for: remote fields, low-connectivity environments, real-time field decisions
What farmers said
After Year 1, participating farmers and advisors shared their experience with both the Cloud and Edge modes of the drone scouting system.
Easy to adopt, no disruption to existing routines
The Edge mode is rated as very easy to use and faces no adoption resistance. The Cloud mode also met initial expectations with low resistance. Both modes are seen as complementary tools that support existing agronomic practice.
Connectivity shapes the choice
The Cloud mode is constrained by its dependence on stable high-bandwidth connectivity, a significant limitation in rural Belgian fields. The Edge mode eliminates this dependency by performing detection on the drone itself.
Edge strongly preferred -and reduces the digital divide
Users strongly favour the Edge mode on both technical and practical grounds. Local processing keeps data on-site, reduces bandwidth and energy use, and ensures reliable operation. The Edge mode is also perceived as improving accessibility and reducing the digital divide.
“Detect only where needed. Spray only where needed.”
Year 1 headline results
Edge completes a full detection cycle in 1.2 seconds. Cloud takes approximately 27 seconds. Both are accurate – Edge is simply faster.
Edge reduces cost per drone mission from approximately 127 EUR to 94 EUR by eliminating high-volume image uploads.
Cloud: mAP50 0.701. Edge: mAP50 0.687. The gap is small — the beetle is visually distinct enough that even lightweight models detect it reliably.
Year 1 results: full breakdown
Results are organised across 4 dimensions. Each compares the Cloud and Edge deployment mode side by side. Dimension 4 is represented above in the section What farmers said.
| Metric | Cloud | Edge | What this means |
|---|---|---|---|
| Time to detection (TTD) | Cloud: ~40s over stable Wi-Fi. Over 4G/3G, upload time for full-resolution images can add minutes. | Edge: ~1.2s. Only compact result payloads transmitted. Predictable regardless of connectivity. | Edge delivers near-instant detection. Cloud latency is network-dependent — a significant limitation in rural fields. |
| ML model accuracy | Precision: 0.76 / Recall: 0.674 / mAP50: 0.701 | Precision: 0.71 / Recall: 0.675 / mAP50: 0.687 | Cloud marginally higher. Gap is small — the beetle is visually distinct. Edge efficiency outweighs the marginal accuracy gain for real-time deployment. |
| Production cost per mission | Approximately 127 EUR. Communications (15 EUR) and cloud compute (20 EUR) are main variable costs. | Approximately 94 EUR. Communications (2 EUR) and cloud compute (2 EUR) minimal. | Edge saves approximately 26% per mission. One avoided spray application can offset several monitoring flights. |
Note: Production cost figures are illustrative benchmarks. Actual costs vary by region, telecom tariffs, and cloud pricing.
| Metric | Cloud | Edge | What this means |
|---|---|---|---|
| Computing resources | Cloud server sustains 35-37W during inference. CPU at 60-90% load. Full detection cycle: ~27s. | Raspberry Pi 5: 6-13W. AI accelerator handles inference. Full detection cycle: 1.2s. | Edge completes detection 22x faster by offloading to a dedicated AI accelerator. |
| Network utilisation | Sustained ~25,000 kB/s during image upload. Requires reliable high-bandwidth uplink. | Brief low-volume spikes only — compact result payloads. Highly resilient to poor connectivity. | Edge sends only results, not images. Near-zero bandwidth dependency. |
| Energy for data processing | Cloud server: sustained 35-37W plateau. Raspberry Pi (drone-side): 5-11W during upload. | Raspberry Pi: 6-13W. Cloud server draws 10-14W briefly. No sustained plateau. | Edge keeps the server in a lightweight supporting role. |
| Total energy per mission | Same for Cloud and Edge. DJI M350 RTK: typical 10-min flight = 91.7 Wh. Drone propulsion dominates: 500-600W during flight. A 10-minute mission consumes 90-100 Wh. Optimising flight paths has greater impact on sustainability than computing efficiency. | ||
| Metric | Cloud | Edge | What this means |
|---|---|---|---|
| Data formats | Identical for Cloud and Edge. JPEG input. InfluxDB and Postgres for storage. Prescription maps in ISOXML, shapefiles, or GeoTIFF — directly usable by precision spraying equipment. | ||
| Networking standards | JPEG uploads to S3 via HTTPS/TLS. REST API for image and result registration. High bandwidth required. | Local S3 storage. Only compact results sent via REST API. Resilient to low bandwidth. | Both encrypted and authenticated. Edge leverages communication protocols far more efficiently. |
| Software switchability | Same modular design for both modes. Any component replaceable without rewriting the pipeline. Modular micro-service architecture. YOLO, Streamlit, Postgres, InfluxDB all swappable via REST APIs. | ||
| Farm data switchability | Same for Cloud and Edge. Full hardware and software independence. JPEG input from any drone or camera. Prescription maps in ISOXML, shapefiles, GeoTIFF — compatible with any compliant spraying equipment. | ||
| Hardware switchability | Same for both modes. No vendor lock-in. Any drone or camera producing JPEG. Edge inference: Raspberry Pi, Jetson, or equivalent. Cloud: any CPU or GPU server. | ||
OpenAgri OS-based Services used
The following OpenAgri OS-based services are integrated in SIP3:
- Weather Data Service (real-time and historical climate data for infestation risk prediction)
- Pest Management Platform (centralised database for logging, analysing, and mapping pest outbreaks)
- Digital Farm Calendar (logs UAV scouting sessions and tracks infestation patterns over time)
PILOT DETAILS
- SIP Leader: EV ILVO
- Farmer partner: ISP
- Sector: Arable / Potatoes
- Country: Belgium
- Target group: Farmers and advisors (co-creation)