13.06.23 - Advancements in Land Cover Classification: Swiss Team's Promising Results in FLAIR-One Challenge
A Swiss team took part in IGN (Institut national de l'information géographique et forestière)'s recent Flair 1 challenge and has published its results.
As a reminder, FLAIR (French Land cover from Aerospace ImageRy) is a geospatial dataset of France, with the aim of producing large-scale cartography.
The Flair-one challenge involved more than 20 billion pixels annotated on the acquired data. These data were transmitted to different teams with the aim of training AI models in land cover classification.
Adrien Gressin and Antoine Carreaud from HEIG-VD took part in this challenge and published their results. They have also tested their algorithm on SWISSIMAGE (Office fédéral de topographie swisstopo) data in Murten and Fribourg, and the results are promising. You can consult them here: https://lnkd.in/esYZWjhc !
These results can only encourage STDL in its project of high-resolution mapping of open soil areas in the canton of Fribourg.
#data #project #france #training #ai #landcover