Dakota Agros · Flagship Platform by Dakota AI LLC (West Fargo, ND)
From Drone Flight to Breeder-Ready Plot Data.
Turn field-trial orthomosaics into audited stand counts, 50% bloom dates, and canopy metrics—backed by human-in-the-loop review and validated against field notes.
No form required — includes Moorhead_Trial_Plot_Summary.csv, Moorhead_Sample_Detections.geojson (QGIS/ArcGIS ready), and our 2-page Validation Brief PDF.
Built in the Red River Valley for plant breeders, research agronomists, seed companies, and Contract Research Organizations (CROs).
HelioVision field workbenchAudited plot counts ready for ANOVA & BLUPs.
Inspect center-row plant stands, open blooms, and canopy height metrics on real Red River Valley trial plots—then export clean CSV and GeoJSON files.
Choose turnkey plot tables or the custom model workbench.
Most breeding teams and CROs just want clean phenotypic spreadsheets without managing computer-vision pipelines. Imaging teams who want to train custom visual detectors can license the underlying workbench directly.
PATH A · FOR BREEDERS & CROS (MOST POPULAR)
Turnkey Plot Processing
You fly the trial (or send us your orthomosaics). We run the models, manually audit edge cases in Agros, and hand you clean plot-level CSVs and GeoJSONs ready for R, SAS, or Agrobase.
Quoted per trial · Zero software homework for your field crew
Center-row yield stand counts isolated from adjacent-plot border rows
Intra-row gap %, R5 open-bloom counts, and interpolated 50% bloom dates
Photogrammetric canopy height (CHM) and human-audited QA status per plot
Have your own drone pipeline and need to detect a custom visual trait? Use the Agros 4-step workflow to review candidate detections, export training chips, and train custom models.
We compared Agros bloom measurements to ground-rated 50% bloom notes from 879 sunflower plots across three 2021 yield trials at Moorhead, Minnesota. Flights and ground notes were not always taken on the same day.
BLOOM CALLS
86% agreement, 94% on the clearest flight
Using a simple 50%-in-bloom rule, raw drone bloom status matched ground ratings on 86% of 1,758 plot-by-flight calls across three flights, and on 94% of plots for the clearest flight (Aug 5).
BLOOM DATE
Within 0.8 days on average
On the Moorhead Early trial, with flights six days apart, interpolated 50% bloom dates landed within 0.8 days of ground notes on average, and within two days for 92% of 264 plots.
FLIGHT TIMING MATTERS
We plan flights around the bloom window
On the two trials where the key flights were 12 days apart, interpolated dates averaged 2.5 to 3.4 days off. That is why flight spacing is set during scoping, so your flights bracket 50% bloom.
Method: 2021 Moorhead, Moorhead Early and Moorhead Hybrids yield trials (raw, unadjusted drone output); plot bloom percentage from drone imagery compared to days-after-planting ground notes. These are Dakota AI internal validation results, not a published study. Our founder's peer-reviewed sunflower research: Olson, N., Trostle, C., Meyer, R., & Hulke, B. S. (2024). Canopy closure, yield, and quality under heterogeneous plant spacing in sunflower. Agronomy Journal. doi:10.1002/agj2.21655. Results on other crops, traits and trials depend on imagery and label quality and are evaluated during scoping.
Have Unanalyzed Trial Imagery Sitting on a Hard Drive? Send Us One Orthomosaic.
Before budgeting for next season's field trials, test Dakota Agros on archival imagery you already flew. Send us one stitched trial orthomosaic (GeoTIFF) and plot boundary shapefile—plus your ground notes if you want a blind accuracy comparison.
SELF-SERVE PROOF · LIVE IN BROWSER
21 Interactive Workbench Demos & Sample .ZIP
We don't do unpaid custom processing—instead, we put 21 interactive multi-crop trials, 36 demonstration models, and our full Moorhead CSV/GeoJSON/PDF deliverable package directly in your browser so your team can verify our output quality upfront.
Send us one stitched archival trial orthomosaic (GeoTIFF) and plot boundary shapefile. Receive an audited CSV + GeoJSON plot table across the entire trial flight, plus a quantitative accuracy report against your field ratings. Its cost can be applied toward a pilot or seasonal contract.
Multi-crop readiness across northern trials
Crop & Trait Readiness Matrix
Validated against field notes on sunflower. For other crops, our demonstration models show what we can build; each pilot is trained and checked on your own imagery and ground notes.
Crop Group
Target Phenotypic Traits
Deployment Status
Sunflower
V-stage center-row stand count, row gap %, R5 open-bloom count, interpolated 50% bloom date, canopy height (CHM), foliar disease triage Try Live Demo (Moorhead Plots) ↗
Early emergence stand count, canopy closure rate, soybean pod load yield counting, canola silique phenotyping, pea seed viability Try Live Demo (Soybean Pod & Canola) ↗
Per-Plot Managed Processing · Archival Validation · Guided Pilots
Prove the workflow or scale across your trials.
Start with a full-trial validation report on imagery you already have, run a guided pilot, or scope full-season per-plot processing. Every engagement is quoted for your trial before any work starts.
01 / TURNKEY PLOT PROCESSING
Breeder-ready trial plot tables.
Quoted per trial by plot count, flights and traits
For breeding teams, CROs, and commodity researchers who want audited center-row stand counts, bloom progression, and canopy height tables ready for statistical analysis.
Volume-based pricing across 1 to 4+ seasonal flights
Human-in-the-loop QA review on every plot before lock
Middle-row vs. all-row isolation & intra-row gap metrics
Direct CSV & GeoJSON export for R, SAS, Agrobase & BMS
Send this scope and we'll reply by email with a quote
Inspect the live field tools
Flagship interactive workbenches.
Explore saved Red River Valley trial plots, inspect multi-layer orthomosaics and 3D canopy point clouds, or download our sample deliverable archive directly.
01 / FLAGSHIP 2D FIELD WORKBENCH
HelioVision Field Workbench
Inspect multi-date sunflower trial orthomosaics, review plant-stand and open-bloom detections, correct outlines, and regenerate plot-level CSV statistics in your browser.
Inspect photogrammetric LAS/LAZ point clouds, textured meshes, and GeoTIFF canopy & terrain surfaces directly in WebGL without desktop GIS dependencies.
The engine powering turnkey processing (Path B for imaging teams)
How Agros audits every plot and adapts to new traits.
Whether Dakota Agros processes your flights turnkey or your internal remote sensing team uses the workbench directly, every plot count passes through this four-step audited pipeline.
01 / PREPARE
Align imagery & plot grid
Ingest GeoTIFF orthomosaics, canopy height models, and trial plot boundaries; extract candidate plants, blooms, or custom targets.
02 / REVIEW
Human-in-the-loop QA
Inspect crowded rows, separate touching canopies, and eliminate false positives before any plot summary is locked.
03 / EXPORT
Deliver tables & chips
Generate breeder-ready plot CSVs and QGIS/ArcGIS GeoJSONs—plus reviewed training chips when calibrating a new trait or crop.
04 / ITERATE
Calibrate across trials
Verify agreement on held-out plots or ground notes and extend detectors across locations, dates, and varieties.
Before you begin
Questions breeding & trial teams ask.
How does the Winter "Hard-Drive Backlog Rescue" test work?
Start by inspecting our 21 interactive demos in the HelioVision Workbench and downloading the sample .ZIP above. When you want to benchmark against your own imagery, send one stitched GeoTIFF orthomosaic and plot boundary shapefile (plus ground notes for a blind comparison) and we deliver a complete audited CSV + GeoJSON validation report across the entire trial, quoted by email before any work starts.
What file formats do we receive in a Turnkey Plot Processing package?
You receive plot-level and row-level CSV files formatted for R, SAS, Agrobase, or BMS (with center-row vs. all-row stand counts, intra-row gap %, bloom counts, interpolated 50% bloom date, canopy height, and QA flags), georeferenced GeoJSON polygon layers for QGIS/ArcGIS Pro, and a visual QA brief. Download the sample .ZIP at the top of this page to inspect the exact files.
Can Agros handle crops besides sunflower?
Yes. Sunflower stand, R5 bloom, 50% bloom date, and CHM workflows are in production (validated against field notes on 879 plots from three 2021 trials). For canola, soybean, dry bean, pea, corn, sugarbeet, and potato, our demonstration models show the approach; a pilot trains and checks the model on your own imagery and ground notes.
How long does a pilot last?
A pilot covers one scoped project, not a fixed three-month term, yearly subscription, or lifetime access. Before payment, your written proposal defines the dataset, deliverables, delivery schedule, and included support period. Ongoing service, new projects, extra training, and hosted access require a separate agreement; there is no automatic renewal.
Where will my research data go, and who owns it?
Data transfer, isolated processing, retention, and deliverable rights are agreed in your pilot proposal before imagery is shared. You retain 100% exclusive ownership of your raw orthomosaics, plot maps, and exported phenotypic tables. Please use the inquiry form for project scoping only; do not send confidential files through it.
Let's start with your trial
Ready to validate an archival flight or scope next season?
Tell us the crop, approximate plot count, imagery format (GeoTIFF/shapefile), and target traits and we'll reply by email with a scoped quote for a validation report or guided pilot.
Dakota Agros is the flagship agricultural computer-vision platform built by Dakota AI LLC, alongside our Zero-Exposure Enterprise R&D engines.
Founded by Neil Olson, M.S. (dual graduate degrees in Plant Sciences and Data Science; first author of Olson et al., 2024, Agronomy Journal; FAA Part 107 pilot).