Computer Vision, Imaging and Computer Graphics Theory and Applications: 15th International Joint Conference, VISIGRAPP 2020 Valletta, Malta, February 27–29, 2020, Revised Selected Papers

Preface
Organization
Contents
Computer Graphics Theory and Applications
Unified Model and Framework for Interactive Mixed Entity Systems
	1 Introduction
	2 Related Work
		2.1 MR, NUI and the IoT
		2.2 Hybrid Interactions
	3 The Design-Oriented Mixed-Reality Internal Model
		3.1 Mixed Entities
		3.2 Interactions Between Mixed Entities
	4 Design and Implementation DOMIM-Based Framework
	5 Use Case
		5.1 Design
		5.2 Implementation
	6 Conclusion
	References
Skeleton-and-Trackball Interactive Rotation Specification for 3D Scenes
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 Rotation Axis Computation
		3.2 Controlling the Rotation
		3.3 Improvements of Basic Method
	4 Formative Evaluation
	5 Detailed Evaluation: User Study
		5.1 Evaluation Design
		5.2 Evaluation Execution
		5.3 Analysis of Results
	6 Discussion
		6.1 Technical Aspects
		6.2 Usability and Applicability
	7 Conclusion
	References
CSG Tree Extraction from 3D Point Clouds and Meshes Using a Hybrid Approach
	1 Introduction
	2 Related Work
		2.1 Primitive Segmentation and Fitting
		2.2 CSG Extraction
	3 Background
		3.1 Evolutionary Algorithms
		3.2 CSG Trees and Signed Distance Functions
		3.3 The CSG Tree Extraction Problem
	4 CSG Tree Extraction Pipeline
		4.1 Input Preparation
		4.2 Solid Primitive Segmentation and Fitting
		4.3 Solid Primitive Filtering
		4.4 CSG Tree Extraction
		4.5 CSG Tree Optimization
	5 Evaluation
		5.1 Training of the Neural Network for Primitive Type Detection
		5.2 Results of the Full Pipeline
		5.3 Comparison with Related Approaches
	6 Conclusion and Future Work
	References
Human Computer Interaction Theory and Applications
Intention Understanding for Human-Aware Mobile Robots: Comparing Cues and the Effect of Demographics
	1 Introduction
	2 Related Research
		2.1 Intention Anticipation by Humans
		2.2 Robots Communicating Intention
	3 Approach
		3.1 The Robot: Experimental Platform
		3.2 Experimental Conditions
		3.3 Experimental Design
	4 Results
		4.1 Subjective Results
		4.2 Participant’s Movement Intention
		4.3 Predicting the Robot’s Intentions
	5 Discussion
	6 Conclusions
	References
Tracking Eye Movement for Controlling Real-Time Image-Abstraction Techniques
	1 Introduction
		1.1 Temporary Aesthetic Experiences
		1.2 Challenges for Eye Tracking Based Interaction Techniques
		1.3 Combining Eye Tracking and Image Abstraction
	2 Background and Related Work
		2.1 Interactive Image-abstraction Techniques
		2.2 Interaction Techniques Based on Eye Tracking
		2.3 Eye Tracking in Understanding and Creating Digital Art
	3 Eye Tracking and Image-abstraction Techniques
		3.1 System Setup and Hardware
		3.2 Overview of Software System Components
		3.3 Sensor Data Processing
	4 Eye Tracking Interaction for Image-Abstraction
		4.1 Mapping Sensor Data to Levels-of-Control
		4.2 Interaction Context
		4.3 Exemplary Interaction Modes
	5 Implementation Aspects
		5.1 Overview of System Architecture
		5.2 Integration of Interactive Image-abstraction Techniques
		5.3 Analysis and Processing of Eye Tracking Data
	6 Results and Discussion
		6.1 Fundamental Application Scenarios
		6.2 Runtime Performance
		6.3 Observed Effects on Users
		6.4 Potential Future Research Directions
	7 Conclusions
	References
Information Visualization Theory and Applications
Improving Deep Learning Projections by Neighborhood Analysis
	1 Introduction
	2 Related Work
		2.1 Dimensionality Reduction
		2.2 Deep Learning Projections
	3 NNP Evaluation
		3.1 Hyperparameter Space Exploration
		3.2 Quality Measurement
	4 NNP Evaluation Results
		4.1 Regularization
		4.2 Optimizer
		4.3 Data Augmentation
		4.4 Loss Function
		4.5 Network Architecture
	5 Insights from Evaluation
	6 Improving NNP by Neighborhood Analysis
	7 KNNP Evaluation
		7.1 Quality on Training Data
		7.2 Quality on Testing Data
		7.3 Quality as Function of Training Set Size
		7.4 Computational Scalability
		7.5 Projection Scatterplots
	8 Discussion and Conclusions
	References
Scalable Visual Exploration of 3D Shape Databases via Feature Synthesis and Selection
	1 Introduction
	2 Related Work
	3 Feature Selection Method
		3.1 Preprocessing
		3.2 Local Feature Computation
		3.3 Feature Vector Computation
		3.4 Dimensionality Reduction
	4 Applications
		4.1 Optimal Scatterplot Creation
		4.2 Fast Computation of Near-optimal Projection Scatterplot
		4.3 User-Driven Projection Engineering
		4.4 Use-Cases
	5 Feature Learning Method
		5.1 Experiments and Results
		5.2 Computational Performance
	6 Discussion
	7 Conclusion
	References
Visual Analysis of Linked Musicological Data with the musiXplora
	1 Introduction
	2 Related Work
		2.1 BMLO
		2.2 musiXplora
	3 Data and Linkage
		3.1 Musici: Persons
		3.2 Casae: Institutions
		3.3 Loci: Places
		3.4 Baccae: Objects
		3.5 Res: Terms
		3.6 Catalogus: Media and Titles
		3.7 Eventa: Events
	4 Connecting Facets
		4.1 Subsets
	5 Use Cases
		5.1 The Oldest Still Preserved Clavichord
		5.2 Jazz Musicians
		5.3 Organological Inspection of a Viola d\'amore
		5.4 Piano Rolls: A Special Case
	6 Future Planning
		6.1 Facets Availability
		6.2 Connectivity
		6.3 Complexity
		6.4 Integrating Visualizations
		6.5 Localization
		6.6 Musicological Motivated Extensions
	7 Conclusion
	References
A Research-Teaching Guide for Visual Data Analysis in Digital Humanities
	1 Introduction
	2 Terminology
		2.1 Digital Humanities
		2.2 Visual Data Analysis
		2.3 Research-Teaching Nexus
	3 Theoretical Training
		3.1 Block 1: The Big Picture
		3.2 Block 2: Understanding Domain-specific Methodologies
		3.3 Block 3: Understanding Visualization Techniques and Their Value in the Application Context
	4 Practical Training
		4.1 Interdisciplinary Projects
		4.2 Publishing Project Results
	5 Considerations
	6 Conclusion
	References
Coherent Topological Landscapes for Simulation Ensembles
	1 Introduction
	2 Related Work
	3 Background
	4 Coherent Topological Landscapes
		4.1 Distance Metric
		4.2 Creation of Meta Tree
		4.3 Visualization of Changing Landscapes
	5 Results
		5.1 Synthetic Data
		5.2 Pattern Formation in 2D
		5.3 Cavity Flow in 3D
	6 Conclusion and Discussion
	References
Computer Vision Theory and Applications
Efficient Range Sensing Using Imperceptible Structured Light
	1 Introduction
	2 Related Work
		2.1 Projection Mapping Methods
		2.2 Structured Light Projection Mapping
	3 Proposed System
		3.1 Structured Light for Correspondence
		3.2 Imperceptible Pattern Injection
		3.3 Undistortion and Rectification
		3.4 Pattern Selection
		3.5 Disparity, Depth and Visualization
	4 Experimental Results
		4.1 Experimental Setup
		4.2 Depth Reconstruction and Imperceptibility
		4.3 Point Density and Resolution
		4.4 Depth Extraction Accuracy
		4.5 Operating Speed
		4.6 Discussion
	5 Conclusion
	References
Hierarchical Object Detection and Classification Using SSD Multi-Loss
	1 Introduction
	2 Related Work
	3 SSD-ML Detection Model
		3.1 Inference
		3.2 Training
	4 Experimental Results: PASCAL VOC
		4.1 Dataset: PASCAL VOC
		4.2 Evaluation Metrics
		4.3 Training Details
		4.4 Experiment 1: Non-hierarchical Classification
		4.5 Experiment 2: Hierarchical Classification
		4.6 Experiment 3: Less Available Ground Truth for Classification
	5 Experimental Results: Traffic Surveillance
		5.1 Datasets
		5.2 Experiment T1: Vehicle Detection, Effect of Datasets
		5.3 Experiment T2: Vehicle Detection
		5.4 Experiment T3: Vehicle Classification
	6 Discussion
	7 Conclusions
	References
Scene Text Localization Using Lightweight Convolutional Networks
	1 Introduction
	2 Concepts and Related Work
		2.1 Text Localization and Recognition
		2.2 Deep Learning
	3 MobText
		3.1 Characterization of Text Regions with MobilenetV2
		3.2 Detecting Multiple Instances of Text via SSD
		3.3 Learning
	4 Experimental Protocol
		4.1 Datasets
		4.2 Evaluation Metrics
		4.3 Evaluation Protocols
	5 Results and Discussion
		5.1 Comparisons with Baselines
		5.2 Results on Mobile-Oriented Environment
	6 Conclusions and Future Work
	References
Early Stopping for Two-Stream Fusion Applied to Action Recognition
	1 Introduction
	2 Related Work
	3 Fusion Methods for Action Recognition
	4 Experiments
		4.1 Datasets
		4.2 Experimental Details
		4.3 Results
	5 Conclusions
	References
RGB-D Images Based 3D Plant Growth Prediction by Sequential Images-to-Images Translation with Plant Priors
	1 Introduction
	2 Related Work
		2.1 Plant Image Analysis
		2.2 Image-to-Image Translation
		2.3 Sequential Prediction
	3 Proposed Method
		3.1 Overview of Network Architecture
		3.2 Generator
		3.3 Encoder-Decoder LSTM in Generator
		3.4 Discriminator
	4 Evaluation
		4.1 Dataset
		4.2 Networks for Comparison
		4.3 Result of Network Comparison
		4.4 3D Prediction
		4.5 Effectiveness of Two LSTM Layers
	5 Evaluation with Plant Prior
		5.1 Network Architecture
		5.2 Result
		5.3 3D Prediction
	6 Discussion
	7 Conclusion
	References
ConvPoseCNN2: Prediction and Refinement of Dense 6D Object Pose
	1 Introduction
	2 ConvPoseCNN
		2.1 Aggregation of Dense Orientation Predictions
		2.2 Iterative Refinement
	3 Evaluation
		3.1 Dataset
		3.2 Training Procedure
		3.3 Evaluation Metrics
		3.4 Results
	4 Conclusion
	References
Expression Modeling Using Dynamic Kernels for Quantitative Assessment of Facial Paralysis
	1 Introduction
	2 Literature Review
	3 Proposed Methodology
		3.1 Spatio-Temporal Feature Extraction
		3.2 Training of Gaussian Mixture Model (GMM)
		3.3 Dynamic Kernel Modeling
	4 Experimental Results
		4.1 Dataset Description
		4.2 Performance Analysis on Different Dynamic Kernels
		4.3 Comparison with Existing Approaches
		4.4 Quantitative Analysis
	5 Conclusion
	References
Perceptually-Informed No-Reference Image Harmonisation
	1 Introduction
	2 Related Work
		2.1 Image Compositing and Harmonisation
		2.2 Multi-task Learning, Feature Sharing and Attention
		2.3 No More Masks
	3 Two-Stage Model: Methodology
		3.1 Motivation
		3.2 Approach
		3.3 Detector and Harmoniser Models
		3.4 COCO-Exp Dataset
		3.5 Similarity Metrics
		3.6 Evaluation Procedure
	4 Two-Stage Model: Results
	5 Two Stage Model: Discussion
	6 End-to-End Model: Methodology
		6.1 Model Architectures
		6.2 Optimization Details
	7 End-to-End Model: Results
	8 Discussion
	9 Conclusions and Future Work
	References
On-board UAV Pilots Identification in Counter UAV Images
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 Data Acquisition
		3.2 People Detection
		3.3 Skeleton Estimation
		3.4 Classification
	4 Dataset
	5 Experimental Setup and Results
	6 Conclusion
	References
Exploring Tele-Assistance for Cyber-Physical Systems with MAUI
	1 Introduction
	2 Related Work
		2.1 Remote Collaboration
		2.2 Cyber-Physical System Interaction
		2.3 Remote Desktop User Interfaces
	3 Requirement Analysis
		3.1 Analysis of Collaborator Feedback
		3.2 MAUI Design Decisions
	4 Implementation Details
		4.1 Basic Software Design
		4.2 Data Transmission
		4.3 Web Based User-Interface in Unity
		4.4 Remote UI Orchestration
		4.5 Backend
	5 Extended User Study
		5.1 Evaluated Conditions
		5.2 The Repair Task
		5.3 Experimental Procedure
		5.4 Extended Results
	6 Building a Mobile UI
		6.1 Use Case
		6.2 Developers
		6.3 Handheld AR
	7 Discussion and Conclusion
	References
On the Use of 3D CNNs for Video Saliency Modeling
	1 Introduction
	2 Computational Models for Saliency Prediction
		2.1 Static Saliency Models
		2.2 Dynamic Saliency Models
		2.3 Saliency Prediction for 360 Video/Image
	3 3DSal Model
		3.1 Overview
		3.2 Cuboid Weights Tuning
		3.3 3D CNN Architecture
	4 Experiments and Results
		4.1 Experimental Setup
		4.2 Evaluation
		4.3 360 Adaptation
		4.4 Some Considerations Around Low Level Attention
	5 Conclusion
	References
CNN-Based Deblurring of THz Time-Domain Images
	1 Introduction
		1.1 Related Work
		1.2 Outline
	2 Beam Shape Effects
	3 Deblurring Based on CNN
		3.1 THzNet-2D
		3.2 THzNet-HS
	4 Experiments
		4.1 Creating Synthetic Data
		4.2 Results on Synthetic Data with THzNet-2D
		4.3 Results on Synthetic Data with THzNet-HS
	5 Conclusion
	References
Thermal Image Super-Resolution: A Novel Unsupervised Approach
	1 Introduction
	2 Related Work
		2.1 Benchmark Datasets
		2.2 Super-Resolution
	3 Proposed Approach
		3.1 Architecture
		3.2 Datasets
		3.3 Evaluation
	4 Experimental Results
		4.1 Settings
		4.2 Results
	5 Conclusions
	References
Regression-Based 3D Hand Pose Estimation for Human-Robot Interaction
	1 Introduction
	2 Related Works
		2.1 Existing 3D Hand Pose Benchmarks
	3 Proposed Architecture
		3.1 Region of Interest (ROI) Detector
		3.2 Pose Regression Network
	4 Experiments
		4.1 Fast Region of Interest Detection
		4.2 Pose Estimation from Heatmaps and 3D Regression
		4.3 Evaluation in HRI Environment
	5 Conclusion
	References
Detection and Recognition of Barriers in Egocentric Images for Safe Urban Sidewalks
	1 Introduction
	2 Literature Review
		2.1 Wearable Cameras
		2.2 Available Datasets
		2.3 Visual Content Analysis
	3 Barriers Egocentric Dataset
	4 Experimental Results
	5 Conclusions and Future Work
	References
Correction to: Regression-Based 3D Hand Pose Estimation for Human-Robot Interaction
	Correction to: Chapter “Regression-Based 3D Hand Pose Estimation for Human-Robot Interaction” in: K. Bouatouch et al. (Eds.): Computer Vision, Imaging and Computer Graphics Theory and Applications, CCIS 1474, https://doi.org/10.1007/978-3-030-94893-1_24
Author Index