de
en
Close
Detailsuche
Bibliotheken
Projekt
Imprint
Privacy Policy
de
en
Close
Imprint
Privacy Policy
jump to main content
Search Details
Quicksearch:
OK
Result-List
Title
Title
Content
Content
Page
Page
Search the document
Uncertainty calibration and its application to object detection / von Fabian Thomas Küppers. Mülheim an der Ruhr, Januar 2023
Content
List of Figures
List of Tables
List of Acronyms
Nomenclature
1 Introduction
1.1 Research Question and Novelty
1.2 Structure of this Work
2 Object Detection and Uncertainty Modeling
2.1 Basics of Neural Networks and Image-based Object Detection
2.1.1 Fully-Connected Neural Networks
2.1.2 Convolutional Neural Network
2.1.3 Architectures for Object Detection
2.2 Different Types of Uncertainty
2.2.1 Modeling of Semantic Confidence
2.2.2 Modeling of Spatial Uncertainty
2.3 Reasons for Unreliable Uncertainty
3 Semantic Confidence Calibration
3.1 Related Work in the Context of Confidence Calibration
3.2 Definitions and Metrics for Confidence Calibration
3.2.1 Classification
3.2.2 Object Detection
3.2.3 Instance Segmentation
3.2.4 Semantic Segmentation
3.3 Multivariate Confidence Calibration
3.3.1 Histogram Binning
3.3.2 Scaling Methods
3.4 Experiments for Semantic Confidence Calibration
3.4.1 Object Detection
3.4.2 Instance Segmentation
3.4.3 Semantic Segmentation
3.5 Conclusion for Semantic Confidence Calibration
4 Bayesian Confidence Calibration
4.1 Epistemic Uncertainty Modeling of Confidence Calibration
4.2 Experiments for Bayesian Confidence Calibration
4.3 Conclusion for Bayesian Confidence Calibration
5 Spatial Uncertainty Calibration
5.1 Related Work in the Context of Spatial Uncertainty Calibration
5.2 Definitions and Metrics for Uncertainty Calibration
5.2.1 Quantile Calibration
5.2.2 Distribution Calibration
5.2.3 Variance Calibration
5.3 Review of Methods for Regression Uncertainty Calibration
5.3.1 Non-Parametric Calibration
5.3.2 Parametric Calibration
5.4 Parametric Calibration using Gaussian Processes
5.4.1 Joint Multivariate Calibration
5.4.2 Correlation Estimation and Recalibration
5.5 Experiments for Spatial Uncertainty Calibration
5.6 Conclusion for Spatial Uncertainty Calibration
6 Application of Calibration to Object Tracking
6.1 Object Tracking by Detection
6.2 Recursive Bayesian Filtering
6.3 Discrete Bayes Filter for Object Existence Estimation
6.4 Kalman Filter for Object Position Tracking
6.5 Track Initialization and Association
6.6 Experiments for Calibration in Object Tracking
6.6.1 Evaluations for intermediate Semantic Confidence Calibration
6.6.2 Evaluations for intermediate Spatial Uncertainty Calibration
6.6.3 Discussion of the Effect of intermediate Calibration
6.7 Conclusion for Calibration in Object Tracking
7 Conclusion
7.1 Summary of Semantic Confidence Calibration
7.2 Summary of Bayesian Confidence Calibration
7.3 Summary of Spatial Uncertainty Calibration
7.4 Final Remarks
Bibliography