๐Ÿ”ฌ Peer-reviewed methodology โ€” Computers & Electronics in Agriculture

The first uncertainty-aware digital twin for 3D plant phenotyping

From point clouds to breeding decisions โ€” with calibrated confidence intervals. Know not just what your genotypes will produce, but how confident that prediction is.

PhenoTwin โ€” Digital Twin Dashboard
LIVE
G7
Thermus โ€” 33.7gยฑ 3.7
G10
Ismena โ€” 32.7gยฑ 3.5
G9
Annika โ€” 31.7gยฑ 4.1
G3
Kinnan โ€” 31.2gยฑ 2.2
G17
Heder โ€” 28.2gยฑ 4.2
G16
Tiril โ€” 20.2gยฑ 2.9
๐ŸŸข SELECTION READY โ€” r = 0.836 at 60% of season (p = 0.002)
r = 0.836
Prediction accuracy at 60% season
95%
Calibrated confidence intervals
15
ML models with auto-recommendation
p = 0.002
Permutation-validated significance

Three layers of intelligence

Not just visualization โ€” prediction with honest uncertainty and actionable decision support.

Layer 1
๐Ÿ“ก

Mirror โ€” What's happening now

Live 3D point cloud visualization, trait time-series, replicate consistency, and automatic quality control. See every genotype's current state at a glance.

Any 3D Sensor
Layer 2
๐Ÿ”ฎ

Predict โ€” What will happen

Gaussian Process growth models with calibrated 95% confidence intervals per genotype. Uncertainty widens where data is sparse, narrows where it's dense. Honest predictions.

Novel โ€” GP with Uncertainty
Layer 3
๐Ÿงช

Simulate โ€” What could happen

"What if we stop scanning now?" โ€” Get quantitative answers with selection overlap metrics, traffic-light readiness, and automatic alerts when genotypes deviate from expected trajectories.

Decision Support

Everything from scan to selection

A complete pipeline that transforms 3D sensor data into breeding intelligence.

๐Ÿ—‚๏ธ

3D Point Cloud Viewer

Upload PLY files from PlantEye, LiDAR, or any 3D sensor. Interactive visualization colored by height, NDVI, or custom traits.

๐Ÿค–

15 ML Models

From GP to XGBoost to LSTM. Auto-recommends models for your dataset size and warns about overfitting risk.

๐ŸŽฏ

SHAP Explainability

Know WHY the model predicts. Per-genotype feature attribution verified against ground truth validation.

โฑ๏ธ

Early Selection

Permutation-validated temporal prediction tells you exactly when rankings become reliable. Stop early, save resources.

๐Ÿ“Š

Data Quality Control

Automatic completeness checks, outlier detection, replicate CV% analysis, and PLY file integrity verification.

๐Ÿ“„

Publication Figures

Export at 300/600 DPI in PNG, SVG, or PDF. Download ZIP packages ready for journal submission.

From sensor to decision in minutes

Upload your data. PhenoTwin handles everything else.

๐Ÿ“ก
Upload
PLY + Excel
โ†’
๐Ÿ”
QC
Auto quality control
โ†’
๐Ÿ“ˆ
Model
GP with 95% CI
โ†’
โœ…
Validate
Ground truth
โ†’
๐Ÿšฆ
Decide
Select with confidence

Built for researchers, priced for reality

Start free. Scale when you're ready.

Open Source
Free
forever
  • GP growth curves with 95% CI
  • Basic QC & validation
  • SHAP explainability
  • Command-line interface
  • GitHub source code
  • Community support
View on GitHub
Enterprise
โ‚ฌ499
per month
  • Everything in Professional
  • API access for integration
  • Multi-sensor support
  • Custom model training
  • On-premise deployment
  • Dedicated support + SLA
Contact Sales

Built by researchers, for researchers

Developed at the Norwegian University of Life Sciences (NMBU). Peer-reviewed methodology. Validated on real breeding data.

Ready to see your data differently?

Upload your PlantEye data and get uncertainty-aware predictions in minutes. No installation required.

Try Free Demo Contact Us