Time-of-flight based interior sensing for automotive applications : optimization of neural network based training methods with limited data / von M.Sc. Patrick Weyers. Wuppertal, 2021
Inhalt
- List of Abbreviations
- Introduction
- Fundamentals
- Deep Learning Fundamentals
- Multi-Layer Perceptron
- Convolutional Neural Networks
- Recurrent Neural Networks
- Optimization Through Backpropagation
- Metrics and Test Procedures to Evaluate Classification Systems
- Optical Flow
- Hardware and Environment
- Related Work
- Driver Monitoring
- Deep Learning for Image Classification and Single Image Interior Sensing
- Action Recognition and Driver ActivityRecognition
- Improving Interior Sensing and DriverMonitoring with Hierarchical Classification
- Hierarchical Occupancy and Driver State Data
- Hierarchical Multi-Label Classification
- Driver State Monitoring System
- Driver State Sensing Results
- Summary and Conclusion of the Hierarchical Multi-Label Driver Monitoring
- Suggestion of an Action and Object Interaction Recognition System for Driver Monitoring
- Action and Object Interaction Recognition Data
- Time Augmentation
- Reduced Features
- Body Keypoints
- Hand Crops
- Hand Crop Normalization
- Reduced Feature Dataset
- Reduced Features Action Recognition System
- Reduced Features Action Recognition Results
- Action and Object Interaction Recognition System Summary
- Improving Continuous Online Action Recognition from Short Isolated Sequences
- Isolated Action Data
- Continuous Action Recognition System
- Class Transitions
- State Handling
- Recurrent States
- Recurrent State Handling
- State Memory Network Adaptation
- Recurrent State handling Results
- Continuous Online Action Recognition from Short Isolated Sequences Summary
- Conclusion and Outlook
- Published Work and Patents
- Bibliography
- List of Figures
- List of Tables
- Appendix
- Hierarchical Classification for Interior Sensingand Driver Monitoring
- Action and Object Interaction
- Continuous Action Recognition From ShortIsolated Sequences