Sensor- and Video-Based Activity and Behavior Computing: Proceedings of 3rd International Conference on Activity and Behavior Computing (ABC 2021) (Smart Innovation, Systems and Technologies, 291)

Organizing Committee
Preface
About This Book
Contents
About the Editors
Toward the Analysis of Office Workers\' Mental Indicators Based on Wearable, Work Activity, and Weather Data
	1 Introduction
	2 Related Research
	3 Data Overview
		3.1 Sensor Data
		3.2 Work Task/Environment Data
		3.3 Psychological Measures
	4 Methods
		4.1 Preprocessing and Feature Extraction
		4.2 Model Development
		4.3 Gini Index
		4.4 SHAP Value Analysis and Comparison
	5 Results
		5.1 Data Statistics
		5.2 Analysis 1: Predictive Results of Psychological Indicators
		5.3 Analysis 2: Relationship Between Psychological Indicators and Behavior
	6 Discussion
		6.1 Discussion for Analysis 1
		6.2 Discussion for Analysis 2
	7 Conclusion
	References
Open-Source Data Collection for Activity Studies at Scale
	1 Introduction
	2 Related Work
	3 Our Proposed Approach
	4 Performance Analysis
	5 Conclusions
	References
Using LUPI to Improve Complex Activity Recognition
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 LUPI Classifier(SVM Plus)
		3.2 Ensemble Classifier
	4 Experimental Evaluation
		4.1 Dataset
		4.2 Implementation and Evaluation Metrics
		4.3 Result
	5 Discussion
		5.1 Improvement of Recognition Accuracy by Using Additional Learning Information
		5.2 Deterioration of Recognition Accuracy Due to the Use of Additional Training Information
	6 Conclusion
	References
Attempts Toward Behavior Recognition of the Asian Black Bears Using an Accelerometer
	1 Introduction
	2 Dataset
		2.1 Basic Information
		2.2 Data Labeling
	3 Behavior Recognition Method
		3.1 Windowing
		3.2 Classification Features
		3.3 Classification
		3.4 Label Extension and Oversampling
	4 Experiment
		4.1 Common Settings
		4.2 Experiment: Difference in Classifier
		4.3 Experiment: Individual Dependency and Training Data Type
		4.4 Experiment: Effectiveness of Features
	5 Conclusion
	References
Using Human Body Capacitance Sensing to Monitor Leg Motion Dominated Activities with a Wrist Worn Device
	1 Introduction
		1.1 Related Work
	2 Physical Background and Sensing Prototype
	3 Activity Recognition Exploration
		3.1 Experiment Setup
		3.2 Exercise Classification with RF and DNN
		3.3 Exercise Counting
	4 Limitation and Future Work
	References
BoxerSense: Punch Detection  and Classification Using IMUs
	1 Introduction
	2 Related Work
		2.1 Vision-Based Exercise Recognition
		2.2 Sensor-Based Exercise Recognition
		2.3 Recognition of Movement Repetition Based Exercises
		2.4 Recognition of Fast Movement
		2.5 Boxing Supporting System
		2.6 Remaining Problems of Existing Research
	3 Proposed Methods to Detect and Classify Punches
		3.1 Overview
		3.2 Target Activities
		3.3 Activity Detection
		3.4 Activity Classification
	4 Validation of Proposed Methods
		4.1 Data Collection Method
		4.2 Experiment
		4.3 Activity Detection Result
		4.4 Activity Classification Result
	5 Conclusion and Future Work
	References
FootbSense: Soccer Moves Identification Using a Single IMU
	1 Introduction
	2 Related Works
		2.1 Human Activity Recognition by Accelerometer
		2.2 Recognition of Basic Exercise
		2.3 Recognition of Hand-Motions in Sports
		2.4 Recognition of Foot-Works in Sports
		2.5 Issues and Our Contributions
	3 Proposed Method
		3.1 Definition of Soccer Movements
		3.2 Segmentation of Individual Actions
		3.3 Feature Extraction for Classifier Building
	4 Validation of the Proposed Method
		4.1 Data Collection
		4.2 Validation Results and Discussions for All Six Actions
		4.3 Validation Results and Discussions for Selected Actions
	5 Conclusion and Future Work
	References
A Data-Driven Approach for Online Pre-impact Fall Detection with Wearable Devices
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 Feature Extraction and Reduction
		3.2 Definition of Fall Risk
		3.3 Training of Machine Learning Model
		3.4 Estimation of Fall Risk and Fall Detection with Threshold
	4 Search for Hyperparameters and Regression Models
		4.1 Experimental Setup
		4.2 The Result of Estimation Accuracy
	5 Evaluation
		5.1 Baseline Method
		5.2 Evaluation Metrics
		5.3 Evaluation Results
		5.4 Comparison of Our Proposed with the Baseline
		5.5 The Importance of Features
		5.6 Evaluation of Airbag Activation Time
	6 Conclusion
	References
Modelling Reminder System  for Dementia by Reinforcement Learning
	1 Introduction
	2 Related Work
		2.1 Assistive Technology
		2.2 Reminder System
	3 Method
	4 Experimental Evaluation
		4.1 Data Description
		4.2 Evaluation Method
	5 Result
	6 Discussion and Future Work
	7 Conclusions
	References
Can Ensemble of Classifiers Provide Better Recognition Results in Packaging Activity?
	1 Introduction
	2 Related Work
	3 Dataset
		3.1 Data Collection Setup
		3.2 Dataset Description
		3.3 Dataset Challenges
	4 Methodology
		4.1 Preprocessing
		4.2 Stream and Feature Extraction
		4.3 Model Selection and Post-processing
	5 Results and Analysis
	6 Conclusion
	References
Identification of Food Packaging Activity Using MoCap Sensor Data
	1 Introduction
	2 Dataset Description
	3 Methodology
		3.1 Preprocessing
		3.2 Feature Engineering
		3.3 Classification
	4 Results
	5 Discussion
	6 Conclusion
	References
Lunch-Box Preparation Activity Understanding from Motion Capture Data Using Handcrafted Features
	1 Introduction
	2 Backgrounds
	3 Data Description
	4 Methodology
		4.1 Data Pre-processing
		4.2 Feature Extraction
		4.3 Feature Selection
		4.4 Model Selection and Evaluation
	5 Result and Discussion
	6 Conclusion and Future Works
	7 Appendix
	References
Bento Packaging Activity Recognition Based on Statistical Features
	1 Introduction
	2 Dataset Description
	3 Methodology
		3.1 Data Preprocessing
		3.2 Feature Extraction
	4 Result and Analysis
	5 Conclusion
	References
Using K-Nearest Neighbours Feature Selection for Activity Recognition
	1 Introduction
	2 State of the Art
	3 Materials and Methods
		3.1 Dataset
		3.2 Feature Engineering
		3.3 Bag-of-Words
		3.4 Preprocessing
		3.5 K-Nearest Neighbour Feature Selection
	4 Results
	5 Discussion
	6 Conclusion and Future Work
	References
Bento Packaging Activity Recognition from Motion Capture Data
	1 Introduction
	2 Related Work
	3 Method
		3.1 Dataset
		3.2 Data Prepossessing
		3.3 Feature Extraction
		3.4 Classification and Model Selection
	4 Result Analysis
	5 Conclusion
	6 Appendix
	References
Bento Packaging Activity Recognition with Convolutional LSTM Using Autocorrelation Function  and Majority Vote
	1 Introduction
	2 Challenge
	3 Method
		3.1 Preprocessing
		3.2 Model
		3.3 Loss Function and Optimizer
		3.4 Final Prediction Classes Activation
	4 Evaluation
	5 Conclusion
	References
Summary of the Bento Packaging Activity Recognition Challenge
	1 Introduction
	2 Dataset Specification
		2.1 Details of Activities
		2.2 Experimental Setting
		2.3 Data Format
	3 Challenge Tasks and Results
		3.1 Evaluation Metric
		3.2 Result
	4 Conclusion
	References
Author Index