Activating real precision farming through intelligent technology

  • Svegro

    Opened in 1960
    Sweden

  • AI method

    Image regression

  • Industry

    Farming

For the first step of their AI journey, Svegro wanted to see if they could train an AI model to predict the height of the plants based on photos so that they could instead use cameras to capture this information. Four months later, thanks to a collaboration with Codon Consulting, a company that uses the Peltarion platform, Svegro now has a fully functioning system that uses four cameras to capture the information instead. They have also begun scoping out the next step, precision farming.

02/ Reducing energy consumption in farming

The Swedish herb-growing company Svegro manages over 55,000 square meters of greenhouse space for their plant production - all with just three horticulturalists monitoring the entire greenhouse. They have made it their mission to be as environmentally friendly as possible, and in the last three years, they have cut waste by half and reduced their energy consumption by 20%. 

To take their environmental action to the next level and also increase their profit margins, they started on a journey to explore precision farming using applied deep learning. Svegro found their data science partner through the Swedish government innovation scheme Vinnova, which has a program called Start your AI journey. Helped by funding from the scheme, they then proceeded to get their first project up and running together with Codon Consulting.  

Lisa Lindström (Svegro) and Erik Fredlund (Codon) in the greenhouse

03/ The benefits of precision farming

For Svegro, the demand side of their business is usually quite predictable throughout the year. They know just when they need to have their plants ready for people’s Christmas food shopping, and they have a pretty good idea of how many plants they will need to produce by then to meet demand. The supply side of growing the plants, however, is more variable, and with a changing climate, it has become even more so in recent years. To predict when plants will be ready to be sold, the staff have to go around and measure the height of the plants on a daily basis.

Svegro wanted both a better way to track the growth of their plants to optimise their production timelines, and had two key interest areas that they wanted to explore further:

  1. Optimising their plant production timeline: One of the main challenges for the company is to produce plants that are exactly the right size for a supermarket shelf at the time when they need to be available. If the plants are too small or too large, they can’t be used. 
  2. Saving money on energy: There are significant cost savings to be made if it turns out that plants can be grown with even a slightly reduced energy consumption in terms of heat and light than previously thought.. Greenhouse farming in northern Europe requires adding energy during the dark moths of the year leaving the company with hefty energy costs during this period, so small adjustments here can have a big impact on profit margins. At the moment, however, the company mostly refrains from experimenting with this, as it can be very costly if they lose produce as a result -- and sophisticated measurement methods needs to be developed to catch the test results in time to avoid quality issues.

In short, both interest areas required having a way to track the growth and potential changes to the plants over time without having to rely on human judgement and repetitive work. The first step on the way to doing this required designing a model that could use image data to say how tall each plant was, so that the manual labour of measuring the 55,000 square meters of plants was removed, and the height of each plant would be provided simply through the photos. 

04/ Setting up the equipment

About a quarter of the company’s produce is basil plants, of which they produce around 6 million pots each year, so it made sense to start here. Basil also has the added benefit of very quickly showing signs of how it reacts to the environment. The greenhouse was fitted out with four cameras that captured images of the plants at regular intervals, and to ensure that a plant could be tracked even if it was moved, they added QR codes to some of the plant pots and sticks with centimeter markings to have a way of indicating scale. The space also had sensor equipment that measured temperature, light, carbon dioxide levels and the conductivity of the soil fertiliser in different parts of the greenhouse.

We tried both doing it through Keras and through the platform, but setting it up in Peltarion was so quick that it became the obvious choice for the project.

05/ Put some AI on it

The model was first trained using 3,000 images of the plants and their height. They then re-trained it a month later on 6,000 images and the results were significantly improved. The model’s prediction for how high the plant is in the photo is now very similar to the human measurements for this. Using the model’s prediction for the height of the plant, and then combining this with other data points relating to the plant, Svegro now receives an estimated final height for the plant on the date of planned harvest. They also have dashboards with the results, alongside measurements of carbon dioxide, temperature, light and conductivity levels in the soil. 

Mikael Huss, Senior Data Scientist at Codon, explains that the choice of Peltarion for setting up the model was obvious when they learnt how fast it was. ‘We tried both doing it through Keras and through the platform, but setting it up in Peltarion was so quick that it became the obvious choice for the project.’

Lisa Lindström, head of cultivation at Svegro, says that it has been remarkable to see the types of conclusions that the consultants from Codon have been able to draw based on just the data. ‘Because of the Corona pandemic, they have never actually been to the site, but just by looking at the data they have been able to work out things about the greenhouse that only employees that have worked here for many years know about.’ She adds that even though this part of the project is only the first step, it has already made a big difference in how they plan their work. ‘The work the people at Codon have done with the Peltarion platform enables us to predict the heights of each plant based on a photo, and thereby removes a big chunk of our manual work here at Svegro. We’re very excited about taking this to the next level.’’

06/ Uncovering the secret life of basil

Predicting the height of the plants based on photos is only the starting point of a much longer journey. The thing Lisa is most excited about is being able to tie all the factors that have been at play to one particular plant, to better understand how they can use the tools at their disposal to achieve better results for the plants. ‘There is huge potential here, and we can’t wait to uncover it.’