Shop : Details
Shop
Details
58,80 €ISBN 978-3-8440-9906-5Paperback194 Seiten68 Abbildungen257 g21 x 14,8 cmEnglischDissertation
Februar 2025
Nikhil Kapoor
Improving Robustness of Perception DNNs in Various Domains
In recent years, deep learning has gained tremendous popularity for solving various challenges in highly automated driving, such as environment perception, sensor fusion, motion planning, etc. In this context, semantic segmentation appears to be a vital task that helps identify objects in a scene around the car. However, deep learning models frequently struggle with robustness concerning diverse input distribution shifts, including sensor noise, photometric changes, motion blur, and adversarial attacks. This lack of robustness poses a substantial safety challenge that must be addressed to ensure the reliability and safety of highly automated driving systems.
This thesis focuses on the lack of robustness issue in camera-based semantic segmentation tasks. We begin by benchmarking two state-of-the-art semantic segmentation models for their resilience to input corruptions and adversarial attacks. The benchmarking uncovers robustness issues in both models, with adversarial attacks inflicting more damage than corruptions. We then propose three novel methods to enhance these models’ robustness across a variety of attacks and corruptions simultaneously. Detailed evaluations and research directions are presented for each method and open the door to improved safety and reliability of autonomous systems in complex environments.
This thesis focuses on the lack of robustness issue in camera-based semantic segmentation tasks. We begin by benchmarking two state-of-the-art semantic segmentation models for their resilience to input corruptions and adversarial attacks. The benchmarking uncovers robustness issues in both models, with adversarial attacks inflicting more damage than corruptions. We then propose three novel methods to enhance these models’ robustness across a variety of attacks and corruptions simultaneously. Detailed evaluations and research directions are presented for each method and open the door to improved safety and reliability of autonomous systems in complex environments.
Schlagwörter: robustness; computer vision; adversarial defense; corruption robustness; convolutional neural networks
Mitteilungen aus dem Institut für Nachrichtentechnik der Technischen Universität Braunschweig
Herausgegeben von Prof. Dr.-Ing. U. Reimers, Prof. Dr.-Ing. T. Kürner und Prof. Dr.-Ing. T. Fingscheidt, Braunschweig
Band 84
Weitere Formate
Elektronische Publikation (PDF): 978-3-8440-9897-6
Export bibliographischer Daten
Shaker Verlag GmbH
Am Langen Graben 15a
52353 Düren
Am Langen Graben 15a
52353 Düren
Mo. - Do. 8:00 Uhr bis 16:00 Uhr
Fr. 8:00 Uhr bis 15:00 Uhr
Fr. 8:00 Uhr bis 15:00 Uhr
Kontaktieren Sie uns. Wir helfen Ihnen gerne weiter.
