Jump to content
An Integrated Approach for Traffic Scene Understanding from Monocular Cameras / Titelei/Inhaltsverzeichnis
An Integrated Approach for Traffic Scene Understanding from Monocular Cameras / Titelei/Inhaltsverzeichnis
Contents
Chapter
Expand
|
Collapse
Page
Titelei/Inhaltsverzeichnis
Details
1–7
1 Introduction
1–7
Details
1.1 Motivation
Details
1.2 Outline and contributions
Details
8–19
2 Related Work and Fundamental Background
8–19
Details
2.1 Advances in CNN architectures for image processing
Details
2.2 Traffic scene representations from monocular cameras
Details
2.3 Fundamental principles and general framework
Details
20–27
3 Experimental Setup and Data Acquisition
20–27
Details
3.1 Outline of the camera system and test platforms
Details
3.2 Inferring scene points from image space measurements
Details
28–37
4 Network Architecture for Multi-task Feature Sharing
28–37
Details
4.1 General design considerations
Details
4.2 Multi-task learning and architectural implications
Details
4.3 Comparison and choice of the feature encoder architecture
Details
38–49
5 Global Road Topology from Scene Context Recognition
38–49
Details
5.1 Use and taxonomies of the traffic scene context
Details
5.2 Recognition decoder and architecture integration
Details
5.3 Road-topology recognition experiments
Details
50–63
6 Drivable Road Area from Semantic Image Segmentation
50–63
Details
6.1 Traffic scene segmentation as dense classification
Details
6.2 Segmentation decoder architecture and spatial priors
Details
6.3 Experiments on drivable road area segmentation
Details
64–83
7 Road Users from Bounding Box Detection
64–83
Details
7.1 Classification and localization of 2D bounding boxes
Details
7.2 Auxiliary regressands and decoder architecture for spatial reconstruction
Details
7.3 Object detection and reconstruction experiments
Details
84–96
8 Multi-task Integration and Conclusive Experimental Analysis
84–96
Details
8.1 Multi-task decoder and architecture integration
Details
8.2 Practical strategy for the joint training of all perceptual tasks
Details
8.3 Experimental results and comparison
Details
97–99
9 Summary, Conclusion, and Outlook
97–99
Details
100–114
A Appendix
100–114
Details
A.1 Road topology dataset statistics
Details
A.2 Technical specifications of the camera system
Details
A.3 Single-task pre-rec curves for all road topologies
Details
A.4 Overview of the segmentation decoder with Hadamard layer
Details
A.5 Detailed breakdown of the single-task KITTI road segmentation results
Details
A.6 Overview of the SSD decoder with auxiliary regressands
Details
A.7 Dual-task Rec+Seg pre-rec curves for road topology recognition
Details
A.8 Dual-task Rec+Det pre-rec curves for road topology recognition
Details
A.9 Multi-task pre-rec curves for road topology recognition
Details
A.10 Dual-task road topology confusion matrices
Details
A.11 Detailed breakdown of the multi-task KITTI road segmentation results
Details
A.12 Full runtime measurement data
Details
115–142
Bibliography
115–142
Details
Durchsuchen Sie das Werk
Geben Sie ein Keyword in die Suchleiste ein
Partial access
An Integrated Approach for Traffic Scene Understanding from Monocular Cameras
Titelei/Inhaltsverzeichnis
Autoren
Malte Oeljeklaus
DOI
doi.org/10.51202/9783186815125-I
ISBN print: 978-3-18-381512-8
ISBN online: 978-3-18-681512-5
Chapter Preview
Share
Download PDF
Download citation
RIS
BibTeX
Copy DOI link
doi.org/10.51202/9783186815125-I
Share by email
Video schließen
Share by email VDI Verlag eLibrary
Recipient*
Sender*
Message*
Your name
Send message
This site is protected by reCAPTCHA and the Google
Privacy Policy
and
Terms of Service
apply.