AI is the engine. Here is how it works.
No mystery and no name-dropping. This is what our models see, how they decide, and the infrastructure that makes plant-level analysis fast and affordable.
Computer vision that reads a canopy plant by plant.
Verdyn's core is a set of vision models trained on field imagery to recognise the visual signatures of crop stress.
What the AI sees
Drone and in-field photographs of crops and canopy — RGB imagery captured from above and at ground level.
What it detects
Disease signatures, nutrient deficiency, water stress, and pest damage at the plant and row level.
How it decides
Image-classification and segmentation models identify affected plants; a change-detection layer tracks how stress spreads across repeat flights over time.
Why it improves
Every labelled and confirmed field strengthens the models — accuracy compounds as our dataset grows.
From a flight to a decision.
Capture
Drone or phone imagery from the field.
Pre-process
Align, tile, and clean imagery.
Vision models
Classify & segment affected plants.
Per-plant map
Localised stress + guidance.
Dashboard
Field teams review and act.
Built on AWS, accelerated with NVIDIA.
We use industry-standard cloud and GPU infrastructure so analysis is fast, secure, and affordable per acre.
How we use AWS
- Store imagery securely in Amazon S3.
- Run our APIs and processing on AWS Lambda and ECS.
- Host detection and reporting data in Amazon RDS.
- Serve the app and site via CloudFront.
- Train and serve models with Amazon SageMaker.
- Monitor everything with CloudWatch.
- Planned: Amazon Bedrock for future guidance-generation workflows.
How we use NVIDIA
- Train vision models on large sets of field imagery using NVIDIA GPUs.
- Run fast batch inference with TensorRT and Triton Inference Server, so a full flight is analysed quickly.
- A path to NVIDIA Jetson edge devices for on-site or in-drone processing where connectivity is poor.
AWS and NVIDIA are trademarks of their respective owners. Verdyn is not affiliated with, sponsored by, or endorsed by them.
The gap finally closed.
Affordable drones and good phone cameras, cheap GPU inference, and rising pressure on yields and input costs have opened a gap between "organisations need plant-level monitoring across the farms they serve" and "an affordable tool exists." Verdyn fills it.
Who we serve.
Farm organisations — cooperatives, estates, and commercial farm groups — and agricultural NGOs running extension and food-security programmes. Revenue is a per-acre organisational subscription plus custom programme licensing. Launch markets [confirm]. We don't publish an invented total market size.
Where we honestly are.
- MVP detection in development.
- Piloting with selected partner organisations [region].
- Building a labelled field-imagery dataset.
- Waitlist open.
What we are building next.
| Phase | Focus | Status |
|---|---|---|
| 1 | MVP detection on key crops | In progress |
| 2 | Pilot programme & labelled dataset | In progress |
| 3 | Expanded crop & disease coverage | Planned |
| 4 | Change-detection over time | Planned |
| 5 | Edge & offline processing (Jetson) | Exploring |
| 6 | New-market expansion | Future |