In reality, in most of our work, field data and the results of modeling or automated analysis are often closely combined.
There is a research project we are currently working on that is a perfect illustration of this.
Launched in 2024 and named Polymor-FENCE, it aims to develop tools and methods to analyze and anticipate the impacts of fences on animal species and their movements.
Until last spring, one of the first phases of the project focused on the automatic detection of fences from aerial images, using AI models. The goal was to map their presence across vast territories, which is not reasonably possible to do in the field.
However, in order to precisely locate certain fences and, above all, determine their type, on-site surveys are essential. The nature of the materials, height, mesh characteristics, and ground anchoring system: all these parameters must be taken into account in order to correctly analyze the impact of a barrier or fence on species, and must therefore be accurately recorded.
Since this summer, and for a few more weeks to come, we have been traveling the roads and paths of several areas of France, cataloging and photographing the various fences we find, particularly those surrounding parks and photovoltaic power plants (which are at the heart of the Polymor-FENCE project).
This fieldwork, led by Lucie Gendron, focuses on a few dozen areas that have been previously identified as having high conservation value and where the presence of fences is likely to have significant negative effects.
What will this data be used for? We will use it in particular for modeling the impacts of fences. The objective of this phase is to simulate various scenarios, with different types of fencing, for example, and to compare their expected effects on ecological continuity.
This proves that there is no need to choose between using digital technologies and collecting data in the field.