Beschreibung
This dissertation explores the application of deep learning and computer vision to enhance object detection for autonomous driving. The research introduces novel approaches to domain adaptation and out-of-distribution detection. Key contributions include adapting image-to-image transfer techniques for object detection, refining these methods to focus on relevant foreground objects, and integrating semi-supervised learning through a student-teacher paradigm. Furthermore, the work proposes active domain adaptation strategies to identify and leverage the most informative data samples. The effectiveness of these combined methods is validated through comprehensive experimental evaluations on automotive benchmark datasets, demonstrating their value in improving object detection systems.
Herstellerkennzeichnung:
Shaker Verlag GmbH
Am Langen Graben 15a
52353 Düren
DE
E-Mail: info@shaker.de
