Case Study

Computer vision and machine learning for carbon measurement

Data Analytics
Computer Vision, ML


The widely adopted measurement techniques of natural capital, including carbon sequestration, are flawed. They are based on outdated science, and rudimentary methodologies authored in the 1960s. This means that wild inaccuracies can exist, both in terms of the number and volume of trees and foliage, and the amount of carbon they sequest. 

We were tasked with using the latest data science and machine learning techniques to the measurement of the number and volume of trees and foliage, thereby contributing to a more accurate, data-driven and trustworthy natural capital product. 


MDRx reviewed the available datasets, which included public satellite images that could be augmented by drone footage. This corpus would be used to train an advanced machine learning engine constituting the product’s core IP. We architected a computer vision solution that would ingest such data for new sites, and provide detailed data analytics. 

We led the product strategy, beginning in Discovery, focussing on our initial target users – large landowners, managers and agents based in the UK. We then geared all of our engineering towards satisfying their needs with tight feedback and iteration loops enabled by Agile Scrum. We developed a land management tool that allowed users to import their GeoJSON land files, or alternatively map them on the platform using Mapbox.  

Our user research enabled us to prioritise a scenario planning feature, whereby users could use predictive analytics to make better, data-driven land management decisions. 

In parallel, the client undertook a complex research project in partnership with leading academics at the University of Oxford and the University of Antwerp, focussed on improving the science that underpins the carbon sequestration calculation. 


  • Analysis through computer vision – the ability to ingest and analyse land images in seconds. 
  • Machine learning insights – a deeper, more data driven understanding of the natural capital associated with land assets 
  • User centricity – a clear and focussed product strategy geared towards solving and addressing user needs, including a predictive scenario planning tool 
  • A successful fundraising – our client successfully raised £2.5m in private sector fundraising, on the back of our work 
  • A better, fairer, more accurate natural capital market – essential ingredients in our global mission to reduce the impacts of climate change and pursue a sustainable future. 

Want to work with us?