As autonomous-driving possibilities are becoming more and more common, the importance of correct image recognition and classification is strongly increasing. A car's computer must be able to recognise what the camera is seeing, and distinguish between objects like cars, trucks, or motorcycles, in order to take the correct action.
We focus on developing methods to increase the robustness and trustability of classification algorithms, in particular in the setting of image classification under driving conditions.
This project is part of Exaigon (Explainable AI systems for gradual industry adoption), an international project developing and applying explainable AI methods to satisfy society's requirements for transparency and trustworthiness of machine learning systems.
Partners: NTNU, SINTEF
International collaborators: University of Wisconsin-Madison and University of Melbourne.