Research case study

Drone-based Christmas tree measurement: from research to 20,000 trees

How Lars Lindved used drone photogrammetry, computer vision and elevation models to count Christmas trees and estimate their height, first in an SDU bachelor project in 2020 and later at field scale in Denmark.

By Lars Lindved

2020

SDU bachelor project

98%+

Hit rate and precision in test data

6.98 cm

RMSE for top-whorl height, 10 trees

20,000

Trees measured in Denmark in 2025

Short answer

Yes. In 2020, the project showed that a consumer-grade UAV system and photogrammetry could locate Christmas trees and estimate the height of their top whorls in test fields. The report recorded a hit rate and precision above 98% for tree detection and approximately 7 cm RMSE for height measurements on 10 control trees. In 2025, Lars collaborated with Dansk Drone Netværk in the context of an ESA Space Solutions project connected with Plan2Fly, measuring approximately 20,000 trees in Denmark as part of research data collection.

Why measure Christmas trees from the air?

Christmas tree production requires a reliable inventory. Producers need to know how many trees are in a plantation, how the trees are distributed by height, and which areas should be prioritised in later operations and harvesting.

The traditional inventory is based on manual counting and sampling. Drone photogrammetry can instead cover a complete field with overlapping images and turn those images into an orthomosaic, an elevation model and a geographic overview of individual trees.

What was tested in 2020?

Lars Lindved’s bachelor project at the University of Southern Denmark was titled "Quantifying properties of Christmas trees using Unmanned Aerial Vehicles". The project ran from 2 February to 1 June 2020 as part of the BSc in Robot Systems.

The aim was to test whether consumer-grade equipment could be used to locate, count and measure Christmas trees. The work was connected to the research context of the "Yield estimation of Christmas trees" project, which SDU describes as a collaboration with RoboCluster, Talmo and Green Team Group. The Odense Robotics article also names Hermes Traffic Intelligence as a project participant.

The workflow

  1. 1. Image acquisition

    A DJI Phantom 4 Pro flew a line or grid pattern and captured overlapping images of the plantation.

  2. 2. Photogrammetry

    The images were processed into an orthomosaic, a digital surface model and a point cloud. Pix4Dmapper, Agisoft Metashape and OpenDroneMap were compared.

  3. 3. Elevation model

    A DSM was normalised into a CHM using terrain points and QGIS, allowing vegetation height to be analysed relative to the ground.

  4. 4. Tree detection

    Python and MATLAB methods used grayscale morphology and local maxima to find tree locations and heights.

  5. 5. Control and statistics

    The outputs were compared with manual counts and physical measurements of selected trees, followed by height maps and field statistics.

Results from the bachelor project

The key figures below come from Lars Lindved’s report. They describe specific test fields and control trees, not a universal accuracy guarantee for every Christmas tree plantation.

Reported results from the 2020 project
TestMethodMeasurementResult
Tree detection in a 0.6 ha fieldPython method3,052 detections, 32 missed trees, 14 false detections98.96% hit rate, 99.54% precision
Height of 10 control treesOpenDroneMap DSM against top whorlR² = 0.9347RMSE = 6.98 cm
Height of 10 control treesOpenDroneMap CHM and PythonR² = 0.944RMSE = 6.34 cm

From proof of concept to 20,000 trees

In 2025, the work moved from a bounded development and test project to larger-scale field data collection. Lars collaborated with Dansk Drone Netværk, a nationwide network of drone pilots, and measured approximately 20,000 Christmas trees in Denmark.

The work took place in the context of an ESA Space Solutions project connected with Dansk Drone Netværk and Plan2Fly. Its purpose was to collect research data about Christmas tree production, not to present the total as an independent certification of every individual measurement.

That is the key progression from 2020 to 2025: first, the method was tested on concrete control fields. Later, the experience was used in an operation covering approximately 20,000 trees in Denmark.

Year
2025
Collaboration
Lars Lindved and Dansk Drone Netværk
Scale
Approximately 20,000 Christmas trees in Denmark
Context
ESA Space Solutions and Plan2Fly

Public context: ESA Space Solutions and Plan2Fly.

What the results do not say

  • The 2020 results are a documented proof of concept, not a statistical comparison with the full population in every plantation.
  • The height measurements were validated on 10 control trees. That is a useful control set, but it is small.
  • Top-whorl height and total tree height are not the same. A method must describe precisely what it measures.
  • The reported 2025 total of approximately 20,000 trees is our project record. The public ESA and industry sources document the project context, but do not publish this aggregate figure themselves.

What does this mean for new projects?

A good drone project starts with a clear measurement question, not a visualisation. Do you need to count trees, estimate height, track growth or find variation in a field? The answer determines flight height, overlap, camera, control measurements and the later analysis.

Today, Forresten and GIJA MAPS work with drone photogrammetry, orthophotos, DSMs, point clouds and specialised analyses. If you want to explore a similar workflow for Christmas trees, forestry, agriculture or another object class, we can help assess the data and the required processing.

Sources and data provenance

This page distinguishes between primary project records and external sources. The 2020 figures are transcribed from the bachelor report. The external links document the research and collaboration context.

Frequently asked questions

What did the 2020 Christmas tree project show?

The project showed that drone photogrammetry with consumer-grade equipment could be used to locate Christmas trees and estimate their top-whorl height in the tested fields. Tree detection had a hit rate and precision above 98%, and OpenDroneMap height measurements for 10 control trees had an RMSE of 6.98 cm.

How accurate was the height measurement?

In an OpenDroneMap test against top-whorl height for 10 control trees, R² was reported as 0.9347 and RMSE as 6.98 cm. A CHM variant with Python gave R² = 0.944 and RMSE = 6.34 cm. These figures apply to the specific test data.

Did you measure total tree height?

Not consistently. The report found that the height peak in the surface models often represented the top whorl or a branch near the top. It is therefore important to distinguish top-whorl height from total tree height.

How many trees were measured in 2025?

Approximately 20,000 Christmas trees were measured in Denmark in 2025 in a collaboration between Lars Lindved and Dansk Drone Netværk, as part of an ESA Space Solutions-related research project connected with Plan2Fly.

Can Forresten help with Christmas tree counting and height measurement?

Yes. Forresten and GIJA MAPS work with drone photogrammetry and can assess orthophotos, DSMs, point clouds and specialised analyses. Contact us to discuss flight data, control measurements and the right analysis method.

Do you have a field to measure?

Start with your drone images, or contact us to discuss a specialised analysis of trees, crops or other objects.

Create an account and upload imagesContact us about a specialised analysis
Drone-based Christmas tree measurement: from research to 20,000 trees | GIJA MAPS