Seminar: Agro Census 2027 challenges logistics and mobilizes data science
Marcelo Maia presents behind the scenes of IBGE’s operations on a national scale
The behind-the-scenes logistics of one of the country’s largest statistical operations were the focus of the academic seminar held on Tuesday (14). This week’s guest speaker was IMPA Tech professor and IBGE (Brazilian Institute of Geography and Statistics) researcher Marcelo Maia, who introduced the students to the challenges behind the Agricultural Census 2027a survey responsible for updating strategic information on Brazilian rural production.
In front of a room full of students from the Bachelor of Mathematics in Technology and Innovation, Maia explained IBGE’s central role in the production of official data and contextualized the difference between sample and census surveys. “In the Census, we seek to portray the totality. In the case of the Agro Census, this means reaching all the agricultural establishments in the country,” he said.
The scale of the operation helps explain its complexity. There are more than 5 million establishments spread over some 8.5 million km², which requires advanced territorial planning, logistics and mapping solutions. In this context, data science and combinatorial optimization tools become essential to ensure efficiency and precision.
According to the researcher, the dynamic nature of the agricultural sector poses additional challenges. “The data is not static, and the very configuration of the territory changes. We need solutions capable of adapting to this reality, always looking for the best alternative among several viable possibilities,” he explained.
During the presentation, Maia showed how different stages of the census operation can be modeled mathematically. The delimitation of agricultural plots, for example, is carried out with the use of computer vision and techniques from deep learning applied to remote sensing images, making it possible to identify land use patterns with greater precision. The definition of collection points involves the distribution of around a thousand operational bases throughout the country, balancing the coverage area and the volume of establishments served.
Another critical point is the planning of the census takers’ routes. The proposal for 2027 is to replace exploratory journeys with pre-optimized routes, built from the integration of geospatial bases and Earth observation data. This process makes it possible to map the rural road network more accurately and organize fieldwork on a large scale.
The last edition of the Agricultural Census was carried out in 2017. A decade later, updating the data is considered fundamental for formulating public policies that are more in line with the reality of the Brazilian countryside. The new edition should also broaden the scope of the survey, with the unprecedented inclusion of information on traditional peoples and communities.

For the students, the seminar offered an opportunity to bring theory and practice closer together. “Seeing these applications motivates me to keep studying and looking for interesting topics,” said Lucca Moulin, a student in the Computer Science program. “Mathematics directly helps to optimize processes, using tools that model real problems and make them easier to solve,” he added.
Data Science student Julia Quiuqui also highlighted the relevance of the presentation, especially in the context of the country’s territorial challenges. “It was very interesting to learn about new applications of science in a geographical context. In a territory as extensive as Brazil, mapping rural establishments can be a major challenge, but with the use of mathematical tools, this process can be significantly improved,” she said.
Maia is also a postdoctoral researcher at the Fluminense Federal University (UFF) Computing Institute, where he completed his undergraduate, master’s and doctoral degrees in Computing, with a sandwich doctorate at the University of Kent in the UK. His work has been recognized in important academic forums, including awards for best doctoral thesis at the Brazilian Operational Research Symposium (2023) and the IEEE Latin American Conference on Computational Intelligence (2022), among other distinctions.
