HomeAboutServicesTechnologyBlogPricingContact Contact us
Crop field rows photographed from above
Technology

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.

AI technology

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.

The pipeline

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.

Technology & infrastructure

Built on AWS, accelerated with NVIDIA.

We use industry-standard cloud and GPU infrastructure so analysis is fast, secure, and affordable per acre.

aws

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.
NVIDIA

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.

Why now

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.

Market

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.

Traction

Where we honestly are.

  • MVP detection in development.
  • Piloting with selected partner organisations [region].
  • Building a labelled field-imagery dataset.
  • Waitlist open.
Roadmap

What we are building next.

PhaseFocusStatus
1MVP detection on key cropsIn progress
2Pilot programme & labelled datasetIn progress
3Expanded crop & disease coveragePlanned
4Change-detection over timePlanned
5Edge & offline processing (Jetson)Exploring
6New-market expansionFuture