Automating plant growth measurement with 3D imaging

MANUFACTURING INDUSTRY

Manual plant measurement has long been a slow, labor-intensive, and often destructive process in agricultural research. In collaboration with a forward-thinking agri-tech company, we developed a fully automated 3D imaging solution that enables researchers to monitor key plant parameters over time – accurately, efficiently, and non-invasively – using only low-cost, commercially available hardware.

BACKGROUND

From manual tracking to scalable automation

A leading provider of intelligent horticultural lighting came to us with a clear challenge: to automate plant growth measurement in a way that was scalable, precise, and non-destructive. Their research relied on tracking parameters like leaf area, biomass, and plant height – data traditionally gathered through invasive, hands-on methods.

CHALLENGE

Balancing precision, simplicity, and cost

Reliable growth data is essential in plant phenotyping, but conventional methods are often slow and destructive. The client needed a solution that could continuously monitor development, extract advanced metrics such as volume and surface area, and operate with affordable, off-the-shelf components. The ultimate goal was a lab-accurate 3D system that combined precision with scalability and cost-efficiency.

Our client came to us with a clear need: to modernize plant growth tracking without compromising accuracy or budget. The challenge was finding a solution that balanced scientific rigor with practical constraints – and that’s exactly where we thrive.

Fredrik Edlund Oldenburg, Client Manager

Knowit.

SOLUTION

A tailored 3D imaging system built from off-the-shelf components

We started by conducting a feasibility study to assess different 3D imaging technologies. While Kinect-based scanning initially seemed like a viable option due to its affordability, it ultimately proved insufficient in terms of accuracy and consistency. As a result, we developed a custom silhouette-based system specifically tailored for precise plant measurement.

The solution consists of a motorized turntable controlled by Arduino, high-definition webcams positioned to capture the plant from multiple angles, and a monitor that alternates between red and blue backgrounds to support effective silhouette extraction. Each plant was photographed from more than 40 different perspectives, and the resulting images were processed using OpenCV and 3DSOM to produce highly accurate 3D models.

Technical overview

The system was composed of modular, off-the-shelf components:

Hardware: A custom-built aluminum rig, rotating platform, HD webcams, and a dynamic background monitor
Software: A tailored image segmentation pipeline using OpenCV, with 3D visualization powered by the Point Cloud Library (PCL)
Extracted metrics: Key plant parameters such as height, width, leaf area, surface area, and convex hull volume – all measured non-destructively

RESULT

Reliable plant data without manual labor

The prototype system delivered high-resolution 3D reconstructions of several plant species, including basil, mint, and chrysanthemum. It enabled automated measurement of key growth metrics with a high degree of accuracy and reliability.

Key outcomes included:

Accurate growth data without manual intervention
Use of affordable, off-the-shelf components
A scalable foundation for future automation in greenhouse and research environments

This project exemplifies how tailored engineering and smart design can solve scientific challenges with surprisingly simple tools. We’ve taken a big step toward democratizing plant phenotyping.

Fredrik Edlund Oldenburg, Client Manager

Knowit.

How the solution contributes to Agenda 2030

Goal 9 – Industry, innovation and infrastructure: Demonstrates how accessible technology and engineering can solve complex problems in cost-effective ways.

Goal 12 – Responsible consumption and production: Enables sustainable farming practices by minimizing waste and supporting non-destructive research methods.