AI Tree Inventory
AI finds the trees a field survey missed, and engineers review every proposal with professional tools before anything reaches the register.
- Ownership
- Built for a client
- Status
- Delivered
- Field
- Computer vision · Geospatial
01 / The problem
Surveying a forest or a city's tree stock by hand is slow and always incomplete. Engineers fly a drone for orthophotos and scan with LiDAR, and then someone has to turn all of it into a register of every tree: species, height, crown, trunk diameter, position.
02 / What we built
7 things
it does well.
- 01
Detects every tree from LiDAR with its height and crown, and matches detections against an existing register to find the missing ones.
- 02
An AI scan of the orthophoto (DeepForest) proposes the trees the first pass missed, each with its confidence and its height from LiDAR.
- 03
Engineers review zone by zone in a map-based app: accept or reject each tree, add what the AI missed, draw or cut canopy by hand.
- 04
Out-of-range measurements are flagged and replaced with real values from nearby trees of the same species, never with an invented number.
- 05
The original register is never modified. Every correction lives in its own log and can be withdrawn.
- 06
Exports to what engineers use: Excel, CSV, GeoJSON, shapefile, GeoPackage and DXF, with crown outlines and the merged canopy.
- 07
Optional QR labels that open a public mobile datasheet for each tree.
03 / Principle
The AI suggests, the engineer decides
Only species from the project's own list are allowed, nothing is ever silently overwritten, and every change is traceable back to the person who approved it.
04 / Screens
In use.
05 / Built with
Engineering note
From orthophoto to a register engineers sign off
The AI proposes trees the field survey missed. Engineers approve, reject and correct on a map, and the result goes out as the client's own Excel register, GIS files, CAD and a QGIS project.
Read the noteNext project
StazhLink