TY - CHAP

T1 - Towards a Camera-Based Road Damage Assessment and Detection for Autonomous Vehicles

T2 - 2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall)

AU - Faßmeyer, Pascal

AU - Kortmann, Felix

AU - Drews, Paul

AU - Funk, Burkhardt

N1 - Conference code: 94

PY - 2021/9/1

Y1 - 2021/9/1

N2 - 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.

AB - 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.

KW - Business informatics

KW - road damage

KW - deep learning

KW - Computer vision

KW - autonomous driving

KW - scaled-YOLOv4

KW - VAE

KW - Wasserstein GAN

UR - http://www.scopus.com/inward/record.url?scp=85122992755&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/1bca900b-296a-33a1-9b3e-333618b4c84b/

U2 - 10.1109/VTC2021-Fall52928.2021.9625213

DO - 10.1109/VTC2021-Fall52928.2021.9625213

M3 - Article in conference proceedings

SN - 978-1-6654-1369-5

T3 - IEEE Vehicular Technology Conference

BT - 2021 IEEE 94th Vehicular Technology Conference, VTC 2021-Fall - Proceedings

PB - IEEE - Institute of Electrical and Electronics Engineers Inc.

CY - Piscataway

Y2 - 27 September 2021 through 28 September 2021

ER -