See every lodged stem from the sky.
CropEye Drone is a crop-monitoring service that flies drones over wheat fields, detects lodging with artificial intelligence, and turns what it sees into precise, practical advice for farmers.
of all the calories and protein consumed worldwide come from wheat.
Source: CGIAR / CIMMYT
When wheat falls over, the harvest falls with it.
Lodging flattens crops
Wind, rain and weak stems or roots can make wheat lean or collapse to the ground.
Lower yield and quality
Flattened plants capture less light, ripen unevenly and are harder to harvest.
Hard to survey on foot
Walking large fields to find and assess damaged areas is slow, costly and inconsistent.
Two kinds of lodging. Two different responses.
CropEye Drone tells the two main types apart, because the cause — and the right advice — is different for each.
The whole plant tilts at the base
The roots lose their anchorage in the soil, so the stem stays straight but leans over from ground level.
The stem bends or breaks
The roots hold, but the stem buckles or snaps part-way up, folding the head of the plant over.
From flight to field advice.
CropEye Drone is delivered as a service: we fly, analyse and advise, so farmers get answers without buying or operating equipment.
- 01 · FLY
Drone survey
A consumer-grade drone captures high-resolution colour images of the field.
RGB camera15 · 45 · 91 m - 02 · SEGMENT
Automatic mapping
Images are automatically segmented to isolate every area of the crop.
Auto segmentation - 03 · DETECT
AI classification
A deep learning model identifies root lodging and stem lodging across the field.
EfficientNetV2-C - 04 · ADVISE
Precision advice
Farmers receive a clear picture of where lodging is, which type it is, and what to do next.
Agronomic guidance
A refined deep learning model, built for the sky.
CropEye Drone's detection engine is based on EfficientNetV2-C, an enhanced version of the EfficientNetV2 image-classification network, with two key additions.
EfficientNetV2-C
An efficient convolutional network that delivers high accuracy with modest computing power — and low carbon emissions.
Coordinate Attention
Helps the network focus on where lodging appears in the image, sharpening the features it learns.
Class-Balanced Focal Loss
Corrects for imbalanced data, so less common lodging patterns are learned as well as the common ones.
The most accurate model at every altitude.
for EfficientNetV2-C in classifying root and stem lodging from drone images.
Best of seven models tested
Outperformed four conventional machine learning models and two other deep learning models at all altitudes.
Minimal carbon emissions
Top accuracy while keeping the computing footprint — and its CO₂ — to a minimum.
Consumer drones are enough
Affordable, consumer-grade drones proved accessible, scalable and sustainable for real-time crop monitoring.
Figures from Dasari et al., IEEE Transactions on Consumer Electronics 71(2), 6924–6932 (2025). doi:10.1109/TCE.2025.3552301
For everyone who depends on a healthy wheat crop.
Wheat farmers
Find damaged areas fast and act on clear, field-specific advice.
Cooperatives
Survey many members' fields with one consistent, objective method.
Agronomists
Base recommendations on aerial evidence instead of spot checks on foot.
Engineering technologists taking farming airborne.
Faculty of Engineering Technologies
CropEye Drone was founded in 2025 by a research team from the Faculty of Engineering Technologies, combining drone operation, image processing and applied AI to deliver precision-agriculture services.
Frequently asked questions
What is wheat lodging?
Lodging is when wheat plants lean or fall over. In root lodging the plant tilts from its base; in stem lodging the stem bends or breaks part-way up.
How accurate is CropEye Drone?
The underlying EfficientNetV2-C model reached an average accuracy of 93.58% in classifying wheat lodging from drone images, the best of the seven models compared.
What kind of drone is used?
Consumer-grade drones with standard high-resolution RGB (colour) cameras. The research showed these are accessible, scalable and sustainable for real-time monitoring.
At what height do the drones fly?
The model was developed and tested on images taken at 15, 45 and 91 metres, and performed best of all models at every one of these altitudes.
Why is it called "low-carbon"?
The model achieves its accuracy with minimal computing-related carbon emissions, so the monitoring itself stays environmentally light.
Published in IEEE Transactions on Consumer Electronics.
Read the peer-reviewed study behind CropEye Drone's wheat-lodging detection.
Dasari, R., Alsaadi, M. A., Nimma, D., Ramesh, J. V. N., Tiwari, M., Jangir, P., Santhi, M. V. B. T., Turayevich, J. I., & Soni, M. (2025). Integration of consumer electronics in precision agriculture using low-carbon wheat lodging drone images. IEEE Transactions on Consumer Electronics, 71(2), 6924–6932.
https://doi.org/10.1109/TCE.2025.3552301