Sensor Data Understanding

Preface 5 I Introduction 11 Fundamental Concept 13 Motivation 15 Active and Assisted Living 16 Digital Medicine 19 Outline and Contribution 21 References 23 II Visual Scene Analysis 27 Large-Scale Multimedia Retrieval 29 Hierarchical Organisation of Semantic Meanings 29 Concept Detection 34 Global versus Local Features 34 Feature Learning 38 Event Retrieval 41 Event Retrieval within Images/Shots 42 Event Retrieval over Shot Sequences 43 Conclusion and Future Trends 44 Reasoning 45 Uncertainties in Concept Detection 45 Adaptive Learning 46 References 49 Shape-Based Object Recognition 63 Problem Statement and Motivation 63 Shape Representation 64 Survey of Related Methods 64 Coarse-grained Shape Representation 67 Fine-grained Shape Representation 68 Shape Matching 71 Survey of Related Methods 71 Shape Matching using Coarse-grained Features 73 Shape Matching using Fine-grained Features 74 Experiments and Results 75 Shape Retrieval using Coarse-grained Features 75 Shape Retrieval using Fine-grained Features 77 Conclusion and Future Trends 78 References 81 Moving Object Analysis for Video Interpretation 91 Object Tracking in 2D Video 91 Survey of Related Approaches 92 Tracking-Learning-Detection 99 Tracking in Omnidirectional Video 100 Experiments and Results 102 3D Trajectory Extraction from 2D Video 103 RJ-MCMC Particle Filtering 104 Convoy Detection in Crowded Video 109 Experiments and Results 112 Conclusion and Future Trends 114 References 116 III Human Data Interpretation 121 Physical Activity Recognition 123 Atomic Activity Recognition 123 Survey of Related Approaches 125 Codebook Approach for Classification 128 Experiments and Results 134 Gait Recognition 139 Survey or Related Approaches 140 Spatiotemporal Representation of Gait 142 Experiments and Results 145 Conclusion and Future Trends 151 References 154 Cognitive Activity Recognition 167 Definition, Taxonomy, Impact on Health 167 Sensing the Brain Activity 168 Electroencephalography 168 Electrooculography 169 Functional Magnetic Resonance Imaging 169 Functional Near-InfraRed Spectroscopy 169 Survey of Related Methods 169 Electrooculography-Based Approach 171 Cognitive Activity Recognition Method 171 Investigating Codewords 172 Application and Validation 173 Collecting a Dataset 173 Implementation Details 175 Results for Cognitive Activity Recognition 175 Results for Codewords Investigation 176 Conclusion and Future Trends 177 References 179 Emotion Recognition 183 Automatic Recognition of Emotions 183 Definition and Taxonomy of Emotions 184 Existing Techniques for Emotion Recognition 190 Emotion Recognition Challenges 192 Multimodal Emotion Recognition 196 Arousal/Valence Estimation 196 Basic Emotion Recognition 199 Approaches Based on Physiological Data 202 Stress Detection Using Hand-crafted Features 204 Codebook Approach for Feature Generation 206 Deep Neural Networks for Feature Generation 209 Conclusion and Future Trends 213 References 214 IV Conclusion 221 Summary and Future Vision 223 Visual Scene Analysis 223 Human Data Interpretation 225 Data-Driven Society 227 References 229 List of Figures 229 List of Tables 233 Pattern Recognition,Machine Learning,Sensor Data Interpretation The rapid development in the area of sensor technology has been responsible for a number of societal phenomena like UGC (User Generated Content) or QS (Quantified Self). Machine learning algorithms benefit a lot from the availability of such huge volumes of digital data. For example, new technical solutions for challenges caused by the demographic change (ageing society) can be proposed in this way, especially in the context of healthcare systems in industrialised countries. The goal of this book is to present selected algorithms for Visual Scene Analysis (VSA, processing UGC) as well as for Human Data Interpretation (HDI, using data produced within the QS movement) and to expose a joint methodological basis between these two scientific directions. While VSA approaches have reached impressive robustness towards human-like interpretation of visual sensor data, HDI methods are still of limited semantic abstraction power. Using selected state-of-the-art examples, this book shows the maturity of approaches towards closing the semantic gap in both areas, VSA and HDI.