An Integrated Approach for Traffic Scene Understanding from Monocular Cameras

Malte Oeljeklaus

doi:10.51202/9783186815125

Zusammenfassung

This thesis investigates methods for traffic scene perception with monocular cameras for a basic environment model in the context of automated vehicles. The developed approach is designed with special attention to the computational limitations present in practical systems. For this purpose, three different scene representations are investigated. These consist of the prevalent road topology as the global scene context, the drivable road area and the detection and spatial reconstruction of other road users. An approach is developed that allows for the simultaneous perception of all environment representations based on a multi-task convolutional neural network. The obtained results demonstrate the efficiency of the multi-task approach. In particular, the effects of shareable image features for the perception of the individual scene representations were found to improve the computational performance.

Contents

Nomenclature VII

1 Introduction 1

1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Outline and contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

2 Related Work and Fundamental Background 8

2.1 Advances in CNN...

Schlagworte

Publikation durchsuchen

Bibliographische Angaben

Copyrightjahr: 2021
Erscheinungsdatum: 28.09.2021
ISBN-Print: 978-3-18-381512-8
ISBN-Online: 978-3-18-681512-5
Verlag: VDI Verlag, Düsseldorf
Reihe: Verkehrstechnik/Fahrzeugtechnik
Band: 815
Sprache: Deutsch
Seiten: 142
Produkttyp: Monographie

Inhaltsverzeichnis

KapitelSeiten

Titelei/Inhaltsverzeichnis Teilzugriff Seiten I - XIV Download Kapitel (PDF)
1. Motivation Kein Zugriff
2. Outline and contributions Kein Zugriff
1. Advances in CNN architectures for image processing Kein Zugriff
2. Traffic scene representations from monocular cameras Kein Zugriff
3. Fundamental principles and general framework Kein Zugriff
1. Outline of the camera system and test platforms Kein Zugriff
2. Inferring scene points from image space measurements Kein Zugriff
1. General design considerations Kein Zugriff
2. Multi-task learning and architectural implications Kein Zugriff
3. Comparison and choice of the feature encoder architecture Kein Zugriff
1. Use and taxonomies of the traffic scene context Kein Zugriff
2. Recognition decoder and architecture integration Kein Zugriff
3. Road-topology recognition experiments Kein Zugriff
1. Traffic scene segmentation as dense classification Kein Zugriff
2. Segmentation decoder architecture and spatial priors Kein Zugriff
3. Experiments on drivable road area segmentation Kein Zugriff
1. Classification and localization of 2D bounding boxes Kein Zugriff
2. Auxiliary regressands and decoder architecture for spatial reconstruction Kein Zugriff
3. Object detection and reconstruction experiments Kein Zugriff
1. Multi-task decoder and architecture integration Kein Zugriff
2. Practical strategy for the joint training of all perceptual tasks Kein Zugriff
3. Experimental results and comparison Kein Zugriff
Summary, Conclusion, and Outlook Kein Zugriff Seiten 97 - 99
1. Road topology dataset statistics Kein Zugriff
2. Technical specifications of the camera system Kein Zugriff
3. Single-task pre-rec curves for all road topologies Kein Zugriff
4. Overview of the segmentation decoder with Hadamard layer Kein Zugriff
5. Detailed breakdown of the single-task KITTI road segmentation results Kein Zugriff
6. Overview of the SSD decoder with auxiliary regressands Kein Zugriff
7. Dual-task Rec+Seg pre-rec curves for road topology recognition Kein Zugriff
8. Dual-task Rec+Det pre-rec curves for road topology recognition Kein Zugriff
9. Multi-task pre-rec curves for road topology recognition Kein Zugriff
10. Dual-task road topology confusion matrices Kein Zugriff
11. Detailed breakdown of the multi-task KITTI road segmentation results Kein Zugriff
12. Full runtime measurement data Kein Zugriff
Bibliography Kein Zugriff Seiten 115 - 142