Human Activity Recognition Challenge

✍ Scribed by Md Atiqur Rahman Ahad, Paula Lago, Sozo Inoue

• Publisher: Springer Singapore
• Year: 2021
• Tongue: English
• Leaves: 133
• Series: Smart Innovation, Systems and Technologies, 199
• Category: Library

✦ Synopsis

The book introduces some challenging methods and solutions to solve the human activity recognition challenge. This book highlights the challenge that will lead the researchers in academia and industry to move further related to human activity recognition and behavior analysis, concentrating on cooking challenge. Current activity recognition systems focus on recognizing either the complex label (macro-activity) or the small steps (micro-activities) but their combined recognition is critical for analysis like the challenge proposed in this book. It has 10 chapters from 13 institutes and 8 countries (Japan, USA, Switzerland, France, Slovenia, China, Bangladesh, and Columbia).

✦ Table of Contents

Preface
Contents
Editors and Contributors
Summary of the Cooking Activity Recognition Challenge
1 Introduction
2 Dataset Description
2.1 Activities Collected
2.2 Experimental Settings and Sensor Modalities
2.3 Data Format
3 Challenge Tasks and Results
3.1 Evaluation Metric
3.2 Results
4 Conclusion
References
Activity Recognition from Skeleton and Acceleration Data Using CNN and GCN
1 Introduction
2 Related Work
3 Method
3.1 GCN for Micro-Activity Recognition
3.2 CNN for Macro-Activity Recognition
4 Experimental Results
4.1 Dataset
4.2 Data Processing
4.3 Results
5 Conclusions
References
Let's Not Make It Complicated—Using Only LightGBM and Naive Bayes for Macro- and Micro-Activity Recognition from a Small Dataset
1 Introduction
2 Challenge and Dataset Details
3 Preprocessing and Feature Engineering
4 Classification Model
5 Experimental Results and Discussion
6 Background and Related Works
7 Conclusion
References
Deep Convolutional Bidirectional LSTM for Complex Activity Recognition with Missing Data
1 Introduction
2 Challenges and Training Pipeline
2.1 Handling Missing Data
2.2 Jitter and Data Augmentation
3 Implemented Deep Learning Architectures
3.1 Convolutional Neural Networks
3.2 LSTM Recurrent Neural Network
3.3 Deep Convolutional Bidirectional LSTM
3.4 Implementation
4 Classification Results
5 Analysis and Discussion
6 Conclusion and Future Work
References
SCAR-Net: Scalable ConvNet for Activity Recognition with Multimodal Sensor Data
1 Introduction
2 Methodology and Result Analysis
2.1 Contribution
2.2 Macro-Activity Recognition
2.3 Micro-Activity Recognition
3 Conclusion
References
Multi-sampling Classifiers for the Cooking Activity Recognition Challenge
1 Introduction
2 Related Work
2.1 Learning Subclass Representations for Visually Varied Image Classification
3 Multi-sampling Classifiers
3.1 Multi-classifiers
3.2 Multi-sampling
4 Experimental Results and Discussion
5 Conclusion
References
Multi-class Multi-label Classification for Cooking Activity Recognition
1 Introduction
2 Related Work
3 Dataset
4 Cooking Activity Recognition Pipeline
5 Results and Discussion
5.1 Macro- and Micro-Activity Recognition Results
5.2 Interpretation of Cooking Activity Features
5.3 Performance with Different Imputation Strategies
5.4 Single-Sensor and Multi-Sensor Feature Performance
6 Limitations and Future Work
7 Conclusions
References
Cooking Activity Recognition with Convolutional LSTM Using Multi-label Loss Function and Majority Vote
1 Introduction
2 Challenge
2.1 Challenge Goal
2.2 Dataset
2.3 Evaluation Criteria
3 Method
3.1 Preprocessing
3.2 Model
3.3 Loss Function and Optimizer
3.4 Final Prediction Classes Activation
4 Evaluation
4.1 Environment
4.2 Result
5 Conclusion
6 Appendix
6.1 Used Sensor Modalities
6.2 Features Used
6.3 Programming Language and Libraries Used
6.4 Window Size and Post-processing
6.5 Training and Testing Time
6.6 Machine Specification
References
Identification of Cooking Preparation Using Motion Capture Data: A Submission to the Cooking Activity Recognition Challenge
1 Introduction
1.1 Method Overview
2 Challenge Data
2.1 Data Preprocessing
3 Feature Extraction
4 Classification
4.1 Classifiers
4.2 Post-Processing with a Hidden Markov Model
5 Results
6 Discussion
7 Conclusion
8 Appendix: General Information
References
Cooking Activity Recognition with Varying Sampling Rates Using Deep Convolutional GRU Framework
1 Introduction
2 Related Works
3 Dataset Description
4 Proposed Methodology
4.1 Preprocessing and Segmentation
4.2 Feature Extraction
4.3 Framework Architecture
5 Results and Discussions
5.1 Macro-Activity Evaluation
5.2 Micro-Activity Evaluation
6 Conclusion
References