Service · Vision-based inventory
Every plant or tree, detected, located and measured
We turn the orthomosaic into a georeferenced inventory: every agave or every canopy with its position, size and row, reviewed on screen by a person. A complete census, not a sample.
What you get
- Inventory per plant or tree with coordinates, diameter (rosette or canopy) and row
- Count per plot, block and fraction, with missing plants and gaps in the row
- History comparable between flights: growth, mortality and replanting
- Export to CSV, GeoJSON and PDF, and access on the web dashboard and in Argos
What we need from you
- An orthomosaic at 5 cm per pixel or better (we produce it, or you send it ready)
- The approximate planting frame (distance between rows and between plants)
- For orchards, the planting date per block if you want age and harvest projection
- Optional: a field count of one fraction to validate against your own reference
How we work
Detection
Computer vision on the orthomosaic: a model trained on millions of agaves for the rosette, and canopy segmentation by vigor and shape for trees.
Review
A person reviews on screen whatever the image leaves in doubt (shadow, closed canopy, weeds). Every correction trains the next model.
Rows and inventory
Plants are grouped into rows, gaps are detected and the result is published per plot, block and fraction, with its history.
Where it is switched on today Proven at scale in agave (over 60 million plants) and in lime orchards. The same method applies to other trees on a regular frame — avocado, orange, pecan, olive — after we evaluate your imagery.
Frequently asked questions
What accuracy can I expect?
In agave on a clean plot the automatic count runs close to 100 % and only drops with weeds or hidden plants; you can check it against the imagery of your own plot. In trees, the on-screen review corrects what a closed canopy or shadow hides. Before quoting we review your imagery and tell you how accurately it can be counted.
Does it work with a closed canopy or touching trees?
Yes, with review. Detection separates canopies by their vigor relief, not just color, and the reviewer settles ambiguous cases. In mature orchards the review effort is higher and we reflect it in the proposal.
Do I need a multispectral drone?
Not for counting. Good-resolution RGB is enough to detect and measure. Multispectral adds the vigor lenses (NDVI, NDRE) and improves the separation between crop and weeds.
Quote this service with your own data
Tell us the area, the crop and which imagery you have. You get a concrete proposal within 48 business hours; and if you like, we first review your imagery and tell you what it can measure.
Or write to us at [email protected]