Towards a Camera-Based Road Damage Assessment and Detection for Autonomous Vehicles: Applying Scaled-YOLO and CVAE-WGAN

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearchpeer-review


Initiatives such as the 2020 IEEE Global Road Damage Detection Challenge prompted extensive research in camera-based road damage detection with Deep Learning, primarily focused on improving the efficiency of road management. However, road damage detection is also relevant for automated driving to optimize passenger comfort and safety. We use the state-of-the-art object detection framework Scaled-YOLOv4 and develop two small-sized models that cope with the limited computational resources in the vehicle. With average F1 scores of 0.54 and 0.586, respectively, the models keep pace with the state-of-the-art solutions of the challenge. Since the data consists only of smartphone images, we also train expert models for autonomous driving utilizing vehicle camera data. In addition to detection, severity assessment is critical. We propose a semi-supervised learning approach based on the encodings learned by combining a class-conditional Variational Autoencoder and a Wasserstein Generative Adversarial Network to classify detected damage into different severity levels.
Original languageEnglish
Title of host publication2021 IEEE 94th Vehicular Technology Conference, VTC 2021-Fall - Proceedings : Proceedings, Virtual Conference, 27 - 30 September 2021
Number of pages7
Place of PublicationPiscataway
PublisherIEEE - Institute of Electrical and Electronics Engineers Inc.
Publication date01.09.2021
ISBN (Print)978-1-6654-1369-5
ISBN (Electronic)978-1-6654-1368-8
Publication statusPublished - 01.09.2021
Event2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall) - Virtuell, Norman, United States
Duration: 27.09.202128.09.2021
Conference number: 94

Bibliographical note

Publisher Copyright:
© 2021 IEEE.

    Research areas

  • Business informatics - road damage, deep learning, Computer vision, autonomous driving, scaled-YOLOv4, VAE, Wasserstein GAN