Technology

Spectroscopy solutions built around calibration.

Our platform starts with a Ventenata prototype and expands to customer-requested plant targets for mixed media or direct plant confirmation.

Why detection is hard

Plant targets vary by medium, moisture, and region.

Subtle signatures

Spectral signals are faint when plants are mixed with other material or environmental noise.

Field variability

Different growth stages and moisture levels can shift spectra in unpredictable ways.

Slow confirmation

Lab methods like DNA testing can be accurate but are not practical for real-time decisions.

Our solution approach

Targeted calibration with repeatable field workflows.

Controlled capture

  • Close-range geometry and internal illumination
  • Smartphone-based workflow for rapid results
  • Designed for mixed media and direct plant scans

Customer-requested targets

  • Ventenata is the prototype calibration
  • Plan to calibrate for any customer-requested plant
  • Field validation tied to real-world samples

The science behind the solutions

NIR spectroscopy and machine learning make fast, field-ready plant confirmation possible.

NIR spectroscopy and plant chemistry

Near-infrared light measures overtone and combination bands tied to O-H, C-H, and N-H bonds. Differences in water, cellulose, lignin, protein, and mineral content create repeatable spectral fingerprints across plant species.

ML detection in mixed media

We use preprocessing and models like PLS-DA, SVM, and lightweight neural networks to isolate species signatures. Training across diverse mixtures enables detection even when targets are diluted in other materials.

Hardware exists, application is new

Miniaturized NIR sensors, stable illumination, and smartphone inference are commercially available and research-backed. To our knowledge, no one has applied them as field-ready solutions for invasive species detection across mixed media and direct plant confirmation.

Prototype plan

Phase I focuses on the Ventenata use case as proof of the calibration platform.

1

Build the labeled dataset

Collect spectra from mixed media and direct plant samples, with metadata for variability.

2

Train and validate

Quantify performance and document failure modes in field conditions.

3

Deploy prototype solutions

Ship handheld units to pilot partners and collect feedback.

Prototype targets (Ventenata)

Clear performance thresholds for Phase I decisions.

Targets

  • At least 95% sensitivity and 90% specificity on held-out samples
  • Stable performance across mixed media and moisture ranges
  • Documented failure modes and mitigation steps

Why it matters

  • Validates the calibration platform for new targets
  • Proves faster-than-DNA confirmation in the field
  • De-risks future customer-requested calibrations

Need a custom calibration solution?

Tell us your target species and sampling context.

Request a consultation