Dual-Mode Training with Style Control and Quality Enhancement for Road Image Domain Adaptation

Venator M, Shen F, Aklanoglu S, Bruns E, Diepold K, Maier A (2020)

Publication Language: English

Publication Type: Conference contribution, Conference Contribution

Publication year: 2020

Publisher: IEEE

Pages Range: 1757-1766

Conference Proceedings Title: 2020 IEEE Winter Conference on Applications of Computer Vision (WACV)

Event location: Snowmass Village, Colorado US

URI: https://ieeexplore.ieee.org/abstract/document/9093401

DOI: 10.1109/WACV45572.2020.9093401

Open Access Link: http://openaccess.thecvf.com/content_WACV_2020/html/Venator_Dual-Mode_Training_with_Style_Control_and_Quality_Enhancement_for_Road_WACV_2020_paper.html

Abstract

Dealing properly with different viewing conditions remains a key challenge for computer vision in autonomous driving. Domain adaptation has opened new possibilities for data augmentation, translating arbitrary road scene images into different environmental conditions. Although multimodal concepts have demonstrated the capability to separate content and style, we find that existing methods fail to reproduce scenes in the exact appearance given by a reference image. In this paper, we address the aforementioned problem by introducing a style alignment loss between output and reference image. We integrate this concept into a multimodal unsupervised image-to-image translation model with a novel dual-mode training process and additional adversarial losses. Focusing on road scene images, we evaluate our model in various aspects including visual quality and feature matching. Our experiments reveal that we are able to significantly improve both style alignment and image quality in different viewing conditions. Adapting concepts from neural style transfer, our new training approach allows to control the output of multimodal domain adaptation, making it possible to generate arbitrary scenes and viewing conditions for data augmentation.

Authors with CRIS profile

Moritz Venator Lehrstuhl für Informatik 5 (Mustererkennung) Andreas Maier Lehrstuhl für Informatik 5 (Mustererkennung)

Involved external institutions

Audi AG DE Germany (DE) Technische Universität München (TUM) DE Germany (DE)

How to cite

APA:

Venator, M., Shen, F., Aklanoglu, S., Bruns, E., Diepold, K., & Maier, A. (2020). Dual-Mode Training with Style Control and Quality Enhancement for Road Image Domain Adaptation. In 2020 IEEE Winter Conference on Applications of Computer Vision (WACV) (pp. 1757-1766). Snowmass Village, Colorado, US: IEEE.

MLA:

Venator, Moritz, et al. "Dual-Mode Training with Style Control and Quality Enhancement for Road Image Domain Adaptation." Proceedings of the The IEEE Winter Conference on Applications of Computer Vision (WACV) 2020, Snowmass Village, Colorado IEEE, 2020. 1757-1766.

BibTeX: Download