Capture
Downward RGB imagery collected during a manually piloted survey flight.
→
The ARIX intelligence pipeline
From aerial imagery
to field-level clarity.
Designed to prove the crop-intelligence workflow first, then move toward autonomous survey and response.
01 / Architecture
A practical system, built in layers.
The initial architecture keeps complex flight autonomy out of the critical path. The pilot flies; the laptop processes; the field map proves the value.
Synchronize
Flight-controller logs align each image with position and time.
→Detect
A dual-track YOLOv8n experiment compares classification and detection approaches.
→Georeference
Detection results become coordinates that can be placed on the field.
→Report
A heat map and PDF translate model output into a practical review layer.
→Illustrative AI detection view02 / Field evidence
Find the signal
inside the crop.
High-resolution imagery gives the model the visual evidence it needs to separate healthy tissue from disease symptoms and place each finding back on the field.
Healthy wheatYellow rustBrown rustSeptoria
03 / Detection concept
Four classes.
One readable map.
The first model targets healthy wheat, yellow rust, brown rust, and Septoria leaf blotch. The final AI approach will be selected after a controlled classification-versus-detection experiment.
Model candidatesYOLOv8n-cls / YOLOv8n-det
ProcessingPost-flight laptop
ValidationAgronomist reviewed
ARIX / FIELD 01 MODEL PREVIEW
·····●·····●·····●·····
Detected classYellow rust
Model confidence91.8%
04 / Product truth
What ARIX is — and what comes next.
Now: a concept-stage manual survey and post-flight crop intelligence pipeline under development.
Next: autonomous waypoint coverage, onboard processing, and targeted spraying after the detection and mapping system is validated.
See the build roadmap →