Computer Vision, Imaging and Computer Graphics Theory and Applications. 16th International Joint Conference, VISIGRAPP 2021 Virtual Event, February 8–10, 2021 Revised Selected Papers

Preface
Organization
Contents
Computer Graphics Theory and Applications
Impact of Avatar Representation in a Virtual Reality-Based Multi-user Tunnel Fire Simulator for Training Purposes
	1 Introduction
	2 Background
	3 FrèjusVR
		3.1 Devices and Technologies
		3.2 Multi-user
		3.3 Avatar Representation Techniques
		3.4 Selected Procedure
	4 Experimental Setup
		4.1 Participants
		4.2 Hardware
		4.3 Methodology
		4.4 Evaluation Criteria
	5 Results and Discussion
		5.1 Embodiment
		5.2 Social Presence
		5.3 Immersion and Presence
		5.4 Direct Comparison
	6 Conclusions and Future Work
	References
Facade Layout Completion with Long Short-Term Memory Networks
	1 Introduction
	2 Related Work
	3 Object Detection and Data Preparation
		3.1 Random Data Generation
		3.2 Object Detection
		3.3 Data Preparation
	4 Recurrent Neural Networks for Pattern Completion and Inpainting Mask Generation
		4.1 Multi-Dimensional Long Short-Term Memory Network
		4.2 Quasi Recurrent Neural Network
		4.3 Rotated Multi-Dimensional Long Short-Term Memory Network
	5 Network Training and Testing
	6 Results
		6.1 Layout Completion
		6.2 Inpainting
		6.3 Limitations
	7 Conclusion and Future Work
	References
Human Computer Interaction Theory and Applications
Generating Haptic Sensations over Spherical Surface
	1 Introduction
	2 Concept and Design of the Spherical Haptic Display
	3 Methods
	4 Results
		4.1 Constructive Wave Interference
		4.2 Combination of Peak Displacement Magnitudes
	5 Providing Enhanced Immersion to Video Content Application Pilot Study
		5.1 Pilot Study Background
		5.2 Experimental Design
	6 Experimental Procedure
	7 Results
		7.1 Pilot Study Discussion
		7.2 Pilot Study Conclusion
	8 Discussion and Future Work
	9 Future Applications
	10 Conclusions
	References
Effects of Emotion-Induction Words on Memory and Pupillary Reactions While Viewing Visual Stimuli with Audio Guide
	1 Introduction
	2 Integration of Visual Information and Auditory Information
		2.1 Memory Formation by Integrating Visual and Auditory Information
		2.2 Emotion-Induction Words for Better Memory
	3 Three Experiments for Investigating Effects of Emotion-Induction Words
		3.1 Measurement Data
		3.2 Word Impression Evaluation Experiment (Exp-W)
		3.3 Sentence Impression Evaluation Experiment (Exp-S)
		3.4 Video Appreciation Experiment (Exp-V)
	4 Results
		4.1 Impression of Emotion-Induction Words
		4.2 Memory Score
	5 Discussion
		5.1 Memory Score Analysis
		5.2 Pupil Diameter Analysis
		5.3 Design Implications
	6 Conclusion and Future Work
	References
A Bimanual Flick-Based Japanese Software Keyboard Using Direct Kanji Input
	1 Introduction
	2 Related Work
	3 Japanese Characters and Keyboards
		3.1 Kana Characters
		3.2 Elements of Kanji Characters
		3.3 Japanese Keyboards
	4 Proposed Method
		4.1 Bimanual Flick
		4.2 Kanji Layouts
		4.3 Direct Kanji Input
		4.4 Key Arrangement
		4.5 Learning Support Functions
	5 Implementation
	6 Comparative Experiments
		6.1 Sentence Input
		6.2 Result of the Sentence Input Experiment
		6.3 Kanji Conversion-Required Word Input
		6.4 Result of the Kanji Conversion-Required Word Input Experiment
		6.5 Subjective Evaluation and Participants\' Comments
	7 Experiment on Learning
		7.1 Method
		7.2 Result
		7.3 Subjective Evaluation and Participants\' Comments
	8 Long-Term Experiment
		8.1 Method
		8.2 Result
	9 Discussion
		9.1 Comparative Experiment
		9.2 Experiment on Learning
		9.3 Long-Term Experiment
		9.4 Direct Kanji Input
	10 Conclusions and Future Work
	References
Comparison of Cardiac Activity and Subjective Measures During Virtual Reality and Real Aircraft Flight
	1 Introduction
	2 Materials and Methods
		2.1 Participants
		2.2 Real Aircraft
		2.3 Virtual Aircraft
		2.4 Flight Scenario
		2.5 Measures
		2.6 Experimental Protocol
		2.7 Data Analyses
	3 Results
		3.1 Flight Parameters
		3.2 Subjective Measures
		3.3 Physiological Measures
	4 Discussion
		4.1 Subjective Measures
		4.2 Cardiac Activity
		4.3 Motion Sickness
		4.4 Conclusion
	References
Information Visualization Theory and Applications
Improving Self-supervised Dimensionality Reduction: Exploring Hyperparameters and Pseudo-Labeling Strategies
	1 Introduction
	2 Background
	3 SSNP Technique
	4 Experimental Setup
		4.1 Datasets
		4.2 Projection Quality Metrics
		4.3 Dimensionality Reduction Techniques Compared Against
		4.4 Clustering Techniques for Pseudo-Labeling
		4.5 Neural Network Hyperparameter Settings
	5 Results
		5.1 Quality on Synthetic Datasets
		5.2 Quality on Real-World Datasets
		5.3 Quality vs Clustering Hyperparameters
		5.4 Quality vs Neural Network Settings
		5.5 Computational Scalability
		5.6 Inverse Projection
		5.7 Data Clustering
		5.8 Implementation Details
	6 Discussion
	7 Conclusion
	References
Visualization of Source Code Similarity Using 2.5D Semantic Software Maps
	1 Introduction
	2 Related Work
	3 Glyph Placement in a 2D Reference Space
		3.1 Data Preprocessing
		3.2 Latent Dirichlet Allocation on Source Code
		3.3 Multidimensional Scaling
	4 Visual Attributes of 3D Glyphs and Use Cases
		4.1 City Metaphor
		4.2 Forest Metaphor
		4.3 Developer Avatars
	5 System Design and Implementation Details
		5.1 Layout Computation
		5.2 Rendering
		5.3 User Interaction
	6 Conclusion
		6.1 Discussion
		6.2 Future Work
	References
Revisiting Order-Preserving, Gap-Avoiding Rectangle Packing
	1 Introduction
	2 The Region Packing Problem
		2.1 Scale Measure
		2.2 Region Alignment
		2.3 Ordering Requirements
	3 The Box Algorithm
	4 The Rectpacking Algorithm
		4.1 ONO Cases
	5 The Improved Rectpacking Algorithm
		5.1 Checking for Stackability
		5.2 Revising the Width
	6 A Constraint-Based Approach
	7 Evaluation
		7.1 Generated Models
		7.2 Real World Models (SCCharts)
		7.3 Hierarchical Graphs
	8 Conclusion and Future Work
	References
Exploratory Data Analysis of Population Level Smartphone-Sensed Data
	1 Introduction
	2 Related Work
		2.1 Health and Context Detection Using Smartphone-Sensed Data
		2.2 Clustering Multivariate Data for Exploratory Data Analysis
		2.3 Visual Analysis of Data Gathered In-the-Wild Using Ubiquitous Devices
	3 Goal and Task Analysis
	4 Our Visual Approach: PLEADES
		4.1 Data Description and Clustering and Dimension Reduction Algorithm Selection and Features View
		4.2 Clustering Results View (CRV)
		4.3 Clusters View (CV)
		4.4 Users View (UV)
		4.5 F-Stats View (FSV)
		4.6 Cluster Details View (CDV)
		4.7 Daily Values View (DVV)
		4.8 Feature Distribution Heatmap (FDH)
		4.9 Saved Results View (SRV)
	5 Evaluation with Use Cases
		5.1 StudentLife (Dataset 1)
		5.2 Use Case 1: Gaining a Quick Overview of the Data
		5.3 Use Case 2: Examining Student Characteristics that Can Be Helpful for Insightful Comparisons
		5.4 Use Case 3: Analyzing Graduate Vs. Undergraduate Students
		5.5 ReadiSens (Dataset 2)
		5.6 Use Case 4: Patterns of Presence in Primary Vs. Secondary Location
		5.7 Use Case 5: Understanding Variances in Behaviors of Potentially Different Shift Workers
	6 Evaluation with Experts
	7 Discussion, Limitations and Future Directions
	8 Conclusion
	References
Towards Interactive Geovisualization Authoring Toolkit for Industry Use Cases
	1 Introduction
	2 Related Work
		2.1 Geovisualization Types
		2.2 Geovisualization Authoring Approaches
		2.3 Limitations of Current Authoring Tools
	3 Geovisto Prototype
	4 Design Requirements Revision
		4.1 Usability
		4.2 Modularity
		4.3 Configurability
		4.4 Extensibility
		4.5 Accessibility
	5 Architecture
		5.1 Core
		5.2 API
	6 Implementation
		6.1 Layers
		6.2 Controls and Utilities
	7 Case Study: Logimic
	8 Discussion
		8.1 Advantages
		8.2 Limitations
	9 Conclusion and Future Work
		9.1 Future Work
	References
Computer Vision Theory and Applications
Global-first Training Strategy with Convolutional Neural Networks to Improve Scale Invariance
	1 Introduction
	2 Background
		2.1 Global Features in Computer Vision
		2.2 Pyramid Based Methods in CNNs for Scale-Invariant Classification
	3 Stacked Filter CNN (SFCNN)
		3.1 Stacked Filters Convolution (SFC) Module
		3.2 Forward and Backward Propagation in SFC Module
	4 Experiments
		4.1 Datasets
		4.2 VGG16 Network
		4.3 SFC Hyper-Parameters
		4.4 The SFCNN Network
		4.5 Training Process
		4.6 Scaled Images for Testing
		4.7 Evaluation Metrics
	5 Results and Discussion
		5.1 Comparing Train and Test Statistics on Non-scaled Images
		5.2 Evaluation of SFCNN on Scaled Images
	6 Conclusion
	References
Spline-Based Dense Medial Descriptors for Image Simplification Using Saliency Maps
	1 Introduction
	2 Related Work
		2.1 Dense Medial Descriptors (DMD)
		2.2 Spline-Based Medial Descriptors (SDMD)
		2.3 Saliency Maps
	3 Proposed 3S-DMD Method
		3.1 User-Driven Saliency Map Generation
		3.2 Saliency-Based Parameter Control
		3.3 Saliency-Aware Quality Metric
	4 Results
		4.1 Evaluation Methodology
		4.2 Effect of Parameters
		4.3 Comparison with DMD and SDMD
		4.4 Comparison with JPEG and JPEG 2000
	5 Discussion
	6 Conclusion
	References
BS-GAENets: Brain-Spatial Feature Learning Via a Graph Deep Autoencoder for Multi-modal Neuroimaging Analysis
	1 Introduction
	2 Existing Work
		2.1 Feature Extraction (FE)
		2.2 Muti-view Representation Learning
		2.3 Graph Representation Learning
	3 Proposed Approach
		3.1 Multi-modal Brain Graphs (BGs) Construction
		3.2 Multi-modal Graph Auto-Encoder Based on the Attention Mechanism (MSCGATE)
		3.3 Trace Regression Predictive Model
	4 Experimental Results
		4.1 InterTVA Data Description
		4.2 Parameters Tuning
		4.3 Performance Evaluation Metrics
		4.4 Prediction Performance
	5 Conclusion
	References
Enhancing Backlight and Spotlight Images by the Retinex-Inspired Bilateral Filter SuPeR-B
	1 Introduction
	2 Background
	3 SuPeR-B
	4 Evaluation
		4.1 Objective Evaluation
		4.2 Subjective Evaluation
		4.3 Comparison
	5 Results
	6 Conclusions
	References
Rethinking RNN-Based Video Object Segmentation
	1 Introduction
	2 Related Work
	3 Method
		3.1 Hybrid Sequence-to-Sequence VOS
		3.2 Bidirectional Architecture
		3.3 Multi-task Training with Optical Flow Prediction
	4 Experiments
		4.1 Implementation Details
		4.2 Experimental Results
		4.3 Ablation on the Impact of Encoder Architecture
		4.4 Ablation on the Impact of RNN Architecture
		4.5 Ablation on Various Designs for the Fusion Block
	5 Conclusion
	References
Author Index