Cassia® is a proprietary technology of Valorhiz. The development of this innovation has benefited from co-funding from the European Regional Development Fund (FEDER, l’Europe s’engage en région Occitanie).
ARTIFICIAL INTELLIGENCE AT THE SERVICE OF ECOSYSTEMS
The innovative Cassia® technology enables the analysis of georeferenced images (from drone, aircraft or satellite acquisition) of vegetation and soil in order to map ecosystems. Based on a learning dataset defined by an expert, the innovative algorithms will analyse the images to learn how to automatically recognize the different entities of interest (habitat, plant community, species, group of plants, soil type …).
To do so, hundreds of pieces of information are extracted from the data and are cross-referenced with each other in the same way that the human eye works. Once the learning process has been successfully completed, the algorithms will be able to apply this approach on very large scales. The maps produced are then validated by experts and enriched with additional data to provide operational tools (dashboard, inventory, risk assessment, etc.).
Follow-up over time
The Cassia® approach follows a standardised method that benefits from a high degree of repeatability, providing reliable and comparable results between different years of observation. This method makes it possible to carry out precise temporal monitoring missions, such as multi-annual monitoring and diachronic comparisons (appearance/disappearance of elements of interest, surface area changes, effects of management actions, vegetation dynamics, etc.).