Overview
The project investigates whether self-supervised vision foundation models (e.g., DINOv3) trained on non-military data can be applied to military image and video data. The aim is to systematically evaluate the performance and robustness of such models in the context of military applications and to develop approaches for improvement.
