AI and machine learning for crop detection. Get to know senior developer Joeri Broeckx
Ten years ago, Joeri Broeckx started his internship at dotNET lab. Today, as a senior developer, he works at the Department of Agriculture and Fisheries. Automatic crop recognition based on AI and machine learning? That is Joeri's cup of IT. What began as a small pilot project in 2019 grew into intelligent and highly efficient crop recognition based on satellite images. A strong achievement that was rewarded with the Agoria eGov Award for Innovation 2020.
IT skills at government level
After studying applied computer science, Joeri knocked on the door of BIT IT Consultancy in 2011, now part of dotNET lab. Programming, improving code quality, learning to work with design patterns: after a training track of several months, Joeri took on his first project as an IT consultant.
The transition from the classroom to the field went very smoothly thanks to dotNET lab. After starting as a junior developer for an agency focused on innovation and technology, I could get started at the Department of Agriculture and Fisheries. And time flies, because that was already seven years ago. Today, I am part of the GIS team, a ten-person team focused on geographic information systems.
Artificial intelligence and machine learning for crop detection
Strong IT talent is scarce and in demand. That is especially true now that innovative technologies such as artificial intelligence are gaining momentum in both the private and public sector. At the Department of Agriculture and Fisheries, the digital revolution is more visible than ever. In 2019, Joeri helped launch an exceptional pioneering project focused on crop recognition through AI and machine learning.
In addition to developing the ASA application, which supports advice for urban planning applications, our team has also focused in recent years on machine learning and AI for crop detection. Europe has specific rules around agricultural subsidies, where farmers receive support based on the type of crop they plant. Of course, inspections are part of that. Those time-consuming field checks, with on-site visits, are now being replaced very efficiently by artificial intelligence. A major step forward.
Growth mindset? A must
Crop recognition happens through machine learning that traces growth curves based on European satellite images and links them to the right crop. Every plant has a unique growth curve. Radar signals allow us to detect it. It is a new technology that requires additional skills, but Joeri was not put off by that. Quite the opposite.
When the project started in 2019, machine learning was completely new to me. So I had to learn. Especially in Python, a popular programming language for this technology, I shifted up several gears. Through dotNET lab's Pluralsight learning platform I mastered the basics, and I learned the rest on the job. Successfully too, because after a few months of testing we saw that the machine learning results closely matched the field inspections.
The versatility of AI
After the successful start period, the department rolled the approach out to other applications. Today, aerial photos are also interpreted automatically to check whether farmers have correctly drawn their plots. According to Joeri, the possibilities are broad.
In AI and machine learning, new developments are constantly being planned. Efficiency is our main focus. For ASA, we keep investing in automation, so administrative overhead stays minimal. That inventive approach was rewarded too: last year, our project won the Agoria eGov Award for Innovation 2020. When you win that prize, you know you are delivering valuable work. Our teamwork could not have asked for a better recognition.