Latest Tweets

Semi-supervised SENtinel-2 TREE Species Detection (SENTREE)

Science [&] Technology Norway (NO)


The objective of the SENTREE (Semi-supervised SENtinel-2 TREE Species Detection) project is to detect tree species in Norwegian production forests by developing deep learning models which will combine high resolution aerial imagery with Sentinel-2 data. A major challenge in utilizing deep learning for tree species detection is the limited amount of training labels and their quality. The SENTREE project will address these issues with semi-supervised learning, noise tolerant training schemes and with automatic label noise detection. The results will be evaluated on a large area covering multiple municipalities in Norway.

Allskog SA is participating in this project as a pilot customer, providing ground truth data, aerial imagery, domain knowledge, and support on validation activities. The project is primed by Science [&] Technology AS and funded by ESA under the EO Science for Society Permanently Open Call funding mechanism


Prime contractor
Science [&] Technology Norway (NO)