Advances in Computer Graphics : 38th Computer Graphics International Conference, CGI 2021 Virtual Event, September 6–10, 2021 Proceedings

Preface
Organization
Contents
Computer Animation
Temporal Parameter-Free Deep Skinning of Animated Meshes
	1 Introduction
	2 Related Work
	3 Temporal Deep Skinning
		3.1 Training and Test Datasets
		3.2 Transformation and Weight Optimization
		3.3 Measuring the Error
		3.4 Building and Tuning a Neural Network for Weight Prediction
	4 Experimental Evaluation of Deep Skinning
		4.1 Quantitative Results
		4.2 Visual Quality Evaluation Results
		4.3 Discussion and Applications
	5 Conclusions
	A Appendix A
	B Appendix B
	C Appendix C
	References
The Impact of Animations in the Perception of a Simulated Crowd
	1 Introduction
	2 Related Work
		2.1 Appearance and Motion of Virtual Humans
		2.2 Crowd Simulation
	3 Experiment Design
		3.1 Stimuli Creation
		3.2 Participants
		3.3 Hypothesis
		3.4 Statistical Analysis
	4 Results
		4.1 Realism of Simulated Crowds (H1)
		4.2 Realism of Trajectories (H2)
		4.3 Realism of Animation (H3)
	5 Discussion
	6 Conclusions and Future Work
	References
Computer Vision
Virtual Haptic System for Shape Recognition Based on Local Curvatures
	1 Introduction
	2 Related Work
	3 Data Capture
		3.1 Stimuli
		3.2 Data Collection
	4 Classifiers
		4.1 Probability Density Function Based
		4.2 Bayesian XGBoost
	5 Results
		5.1 Probability Density Function Based
		5.2 Bayesian XGBoost
	6 Discussion
	7 Conclusions and Future Work
	References
Stable Depth Estimation Within Consecutive Video Frames
	1 Introduction
	2 Related Work
	3 Method
		3.1 Temporal Stability Loss
		3.2 Inconsistency Check and Self-discovered Mask
		3.3 De-scaled Geometry Consistency Loss
		3.4 Network Architecture
	4 Experiments
		4.1 Training Details
		4.2 Comparisons and Ablation Study
	5 Conclusions and Future Work
	References
Progressive Multi-scale Reconstruction for Guided Depth Map Super-Resolution via Deep Residual Gate Fusion Network
	1 Introduction
	2 Related Work
	3 Proposed Method
		3.1 Overview
		3.2 Color Information Extraction Branch
		3.3 Depth Map Super-Resolution Branch
		3.4 Loss Function
	4 Experimental Results
		4.1 Implementation Details
		4.2 Quantitative Evaluation
		4.3 Qualitative Evaluation
	5 Conclusion
	References
SE_EDNet: A Robust Manipulated Faces Detection Algorithm
	1 Introduction
	2 Detection Algorithms
		2.1 Framework
		2.2 Network Structure
		2.3 Image Residuals in YCrCb Color Space
	3 Experiment Analysis
		3.1 Setup
		3.2 Comparison Experiment
		3.3 Robustness Performance Analysis
	4 Conclusion
	References
PointCNN-Based Individual Tree Detection Using LiDAR Point Clouds
	1 Introduction
	2 Method
		2.1 Overview
		2.2 Build CHM
		2.3 Generate Detection Sample
		2.4 Sample Classifier
		2.5 Tree Stagger Analysis
	3 Results
		3.1 Detection Result
		3.2 Comparison with Related Research
	4 Conclusion
	References
Variance Weight Distribution Network Based Noise Sample Learning for Robust Person Re-identification
	1 Introduction
	2 Related Work
		2.1 Deep Person Re-ID Models
		2.2 Person Re-ID with Sample Noise
		2.3 Robust Deep Learning with Label Noise
		2.4 Feature Distribution Modelling
	3 Methodology
		3.1 Conventional Baseline Model
		3.2 Feature Uncertainty Distribution Learning
		3.3 Rectifying Label Learning
		3.4 Overall Classification Loss
	4 Experiments
		4.1 Datasets and Settings
		4.2 Implementation Details
		4.3 Comparison with the State-of-the-Arts
	5 Conclusion
	References
Monocular Dense SLAM with Consistent Deep Depth Prediction
	1 Introduction
	2 Related Work
		2.1 Monocular Visual SLAM
		2.2 Dense Mapping
		2.3 SLAM with Deep Depth Prediction
	3 System Overview
	4 Local Mapping with Depth Refinement
		4.1 2D Image Analysis
		4.2 3D Outlier Detection
	5 Global Dense Mapping with Egomotion Constraints
	6 Evaluation
		6.1 Qualitative Results
		6.2 Quantitative Results
	7 Conclusion
	References
3D Shape-Adapted Garment Generation with Sketches
	1 Introduction
	2 Related Work
	3 The Proposed Method
		3.1 Overview of the Network Architecture
		3.2 Sketch Encoder and Body Shape Encoder
		3.3 Fully Convolutional Mesh Decoder
		3.4 Loss Function
	4 Experiments
		4.1 Dataset Construction
		4.2 Results
	5 Conclusion
	References
Geometric Computing
Light-Weight Multi-view Topology Consistent Facial Geometry and Reflectance Capture
	1 Introduction
	2 Related Work
		2.1 High-Quality Facial Geometry
		2.2 Facial Appearance Capture
	3 System Overview
	4 Proposed Method
		4.1 Landmarks Based Initialization
		4.2 Mesh Deformation
		4.3 Multi-view Based Diffuse-Specular Separation
		4.4 Surface Normal and BRDF Estimation
		4.5 Finer Geometry Optimization
	5 Results
	6 Conclusion
	References
Real-Time Fluid Simulation with Atmospheric Pressure Using Weak Air Particles
	1 Introduction
	2 Related Work
		2.1 Particle-Based Fluid Simulation
		2.2 Fluid Simulation with Atmospheric Pressure
	3 Background
	4 Weak Air Particles
	5 Surface Force Model
		5.1 Density-Related Atmospheric Pressure Force
		5.2 Surface Tension Force
	6 Implementation
	7 Results
	8 Conclusion and Future Work
	References
Human Poses and Gestures
Reinforcement Learning for Quadruped Locomotion
	1 Introduction
		1.1 Objectives
		1.2 Analytic Reviews on Previous Work
	2 Methodology
		2.1 Modelling Quadruped Locomotion
		2.2 Reinforcement Learning
	3 Experiment and Comparative Evaluation
	4 Discussion and Conclusion
	References
Partially Occluded Skeleton Action Recognition Based on Multi-stream Fusion Graph Convolutional Networks
	1 Introduction
	2 Related Work
		2.1 Manual Feature Extraction Method
		2.2 RNN/CNN-Based Method
		2.3 GCN-Based Method
	3 Proposed Method
		3.1 Multimodal Feature Extraction
		3.2 Spatial-Temporal Graph Convolutional Network
		3.3 Occlusion Sensitive Multi-stream Fusion Networks
	4 Experiments
		4.1 Datasets
		4.2 Implementation Details
		4.3 Experimental Results
	5 Conclusion
	References
Social-Scene-Aware Generative Adversarial Networks for Pedestrian Trajectory Prediction
	1 Introduction
	2 Related Work
		2.1 Crowd Interaction
		2.2 Multimodal Trajectory Prediction
	3 Method
		3.1 The Formulation for Pedestrian Trajectory Prediction
		3.2 Scene Module
		3.3 Social Module
		3.4 Generative Adversarial Networks Module
	4 Experiments
		4.1 Evaluation Metrics and Baselines
		4.2 Quantitative Evaluations
		4.3 Qualitative Evaluations
	5 Conclusion
	References
Image Processing
Cecid Fly Defect Detection in Mangoes Using Object Detection Frameworks
	1 Introduction
	2 Towards Automatic Defect Detection in Agricultural Produce
	3 Methodology
		3.1 Image Acquisition
		3.2 Data Preparation
		3.3 Object Detection Frameworks
	4 Experimental Results
	5 Conclusion and Future Works
	References
Twin-Channel Gan: Repair Shape with Twin-Channel Generative Adversarial Network and Structural Constraints*-6pt
	1 Introduction
	2 Related Work
	3 Method
		3.1 Geometry Information Completion
		3.2 Structure Information Optimization
		3.3 Fine-Tune
	4 Experiments and Evaluation
		4.1 Implementation Details
		4.2 Shape Repair
		4.3 Results and Discussion
	5 Limitation and Future Work
	6 Conclusion
	References
CoPaint: Guiding Sketch Painting with Consistent Color and Coherent Generative Adversarial Networks
	1 Introduction
	2 Related Work
		2.1 Generative Adversarial Networks (GANs)
		2.2 Colorization
	3 Methods
		3.1 Overview
		3.2 Dataset
		3.3 Network Structure
		3.4 Loss Function
	4 Experiments
		4.1 Training Strategy
		4.2 Dataset Generation
		4.3 Analysis on Angles
	5 Evaluation
		5.1 Quality Analysis
		5.2 Color Consistency Analysis
		5.3 Ablation Studies
	6 Conclusions and Limitations
	References
Multi-Stream Fusion Network for Multi-Distortion Image Super-Resolution
	1 Introduction
	2 Method
		2.1 Multi-Stream Fusion
		2.2 Fusion Module
		2.3 Deep Supervision
	3 Experimental Results
		3.1 Data Preprocessing and Network Training
		3.2 Model Analysis
		3.3 Results Analysis
	4 Conclusion
	References
Generative Face Parsing Map Guided 3D Face Reconstruction Under Occluded Scenes
	1 Introduction
	2 Related Works
		2.1 Generic Face Reconstruction
		2.2 Face Image Synthesis
	3 Our Approach
		3.1 Landmark Prediction Task
		3.2 Face Parsing Map Generation
		3.3 Face Image Synthesis with GAN
		3.4 Camera and Illumination Model
		3.5 Loss Function of 3D Reconstruction
	4 Implementation Details
	5 Experimental Results
		5.1 Qualitative Comparisons with Recent Works
		5.2 Quantitative Comparison
	6 Conclusions
	References
Compact Double Attention Module Embedded CNN for Palmprint Recognition
	1 Introduction
	2 Related Work
		2.1 CNN-Based Palmprint Recognition Methods
		2.2 Attention Mechanism
	3 The Proposed Method
		3.1 The Framework of CDAM-Net
		3.2 Double Attention Module (DAM)
	4 Experiments
		4.1 Databases
		4.2 Palmprint Identification Results
		4.3 Effectiveness of the DAM
		4.4 Parameter Analysis
	5 Conclusion
	References
M2M: Learning to Enhance Low-Light Image from Model to Mobile FPGA
	1 Introduction
	2 Approach
		2.1 Quantization
		2.2 Hardware Optimizations
	3 Experimental Results
		3.1 Visual Comparisons
		3.2 Comparison with Other Platforms
	4 Conclusion
	References
Character Flow Detection and Rectification for Scene Text Spotting
	1 Introduction
	2 Proposed Method
		2.1 Text Segmentation
		2.2 Text Rectification
		2.3 Text Recognition
	3 Experiments and Discussions
		3.1 Datasets and Implementation Details
		3.2 Text Detection and Rectification Results
		3.3 Comparison with the State of the Art
	4 Conclusions
	References
A Deep Learning Method for 2D Image Stippling
	1 Introduction
	2 Related Work
		2.1 Image Stylization
		2.2 Stippling
		2.3 Deep Point Generation/Processing
	3 Our Method
		3.1 Network Architecture
		3.2 Loss Function
		3.3 Point Removal
	4 Experiment and Training
		4.1 Dataset
		4.2 Network Details
		4.3 Training
		4.4 Other Details
	5 Result
		5.1 Comparison with Previous Works
		5.2 Ablation Experiment
		5.3 Limitations
	6 Conclusion and Future Work
	References
Medical Imaging
In Silico Heart Versatile Graphical Interface with Systole and Diastole Phases Customizable for Diversified Arrhythmias Simulations
	1 Introduction
	2 Heart Simulation Review
	3 Methods
	4 Results
	5 Discussion
	6 Conclusion
	References
ADD-Net:Attention U-Net with Dilated Skip Connection and Dense Connected Decoder for Retinal Vessel Segmentation
	1 Introduction
	2 Related Work
		2.1 Traditional Image Processing
		2.2 Deep Learning
	3 Method
		3.1 Network Structure
		3.2 Attention Module
		3.3 Dilated Skip Connection
		3.4 Dense Connected Decoder
	4 Experiment Results and Analysis
	5 Conclusion
	References
BDFNet: Boundary-Assisted and Discriminative Feature Extraction Network for COVID-19 Lung Infection Segmentation
	1 Introduction
	2 Proposed Method
		2.1 Architecture of BDFNet
		2.2 Joint Loss Functions
	3 Experiments
		3.1 Implementation Details
		3.2 Ablation Study
		3.3 Comparisons with State-of-the-Art Methods
	4 Conclusion
	Appendix
	References
A Classification Network for Ocular Diseases Based on Structure Feature and Visual Attention
	1 Introduction
	2 Material and Method
		2.1 Dataset
		2.2 Date Preprocessing
		2.3 Structure Feature Extraction
		2.4 Classification Network
	3 Results and Discussion
	4 Conclusion
	References
Physics-Based Simulation
DSNet: Dynamic Skin Deformation Prediction by Recurrent Neural Network
	1 Introduction
	2 Dynamic Skin Data and Representation
		2.1 SMPL Model
		2.2 Datasets
		2.3 Generation of Training Data
	3 DSNet
		3.1 Dimension Reduction by an Autoencoder
		3.2 Dynamic Skin Deformation Network
	4 Implementation Details
	5 Results
	6 Conclusion
	References
Curvature Analysis of Sculpted Hair Meshes for Hair Guides Generation
	1 Introduction
	2 Previous Work
	3 Hair Mesh Curvature Analysis
		3.1 Curvature Analysis
		3.2 Generation of Guides
		3.3 Plausibility of Guides
		3.4 Direction Analysis
		3.5 Connection of Guides
		3.6 Repulsion of Guides
		3.7 Vector Field Generation from Guides
		3.8 Scalp Connection
	4 Results
	5 Conclusion
	References
Synthesizing Human Faces Using Latent Space Factorization and Local Weights
	1 Introduction
	2 Related Work
	3 Locally Weighted Autoencoder
		3.1 Pre-Computed Local Weights from NMF
		3.2 Latent Space Manipulation
	4 Implementation
	5 Experimental Results
		5.1 Generation Results
		5.2 Discussion
	6 Conclusion
	References
CFMNet: Coarse-to-Fine Cascaded Feature Mapping Network for Hair Attribute Transfer
	1 Introduction
	2 Related Work
		2.1 GAN-Based Attribute Manipulation
		2.2 3D Hair Reconstruction
	3 Method
		3.1 Overview
		3.2 Predefined Feature Extraction
		3.3 Coarse-to-Fine Cascaded Transfer
	4 Experiments
		4.1 Dataset and Pretraining
		4.2 Result Analysis
		4.3 Comparison
		4.4 Ablation Study
	5 Conclusions
	References
Rendering and Textures
Dynamic Shadow Synthesis Using Silhouette Edge Optimization
	1 Introduction
	2 Related Work
	3 Hash-Culling Approach
	4 Implementation
	5 Results and Discussion
	6 Conclusion
	References
DDISH-GI: Dynamic Distributed Spherical Harmonics Global Illumination
	1 Introduction
	2 Related Work
	3 Distributing SH Light Probe Updating and Usage
		3.1 Path Tracing and SH Probe Update
		3.2 Rasterization and SH Probe Usage
	4 Distributed Computing
	5 Results
		5.1 Quality
		5.2 Performance
	6 Conclusion
	Appendix A GGX Approximation with ZH Lobes
	Appendix B Comparison Images
	References
Simplicity Driven Edge Refinement and Color Reconstruction in Image Vectorization
	1 Introduction
	2 Related Work
		2.1 Curve-Based Design of Vector Graphics
		2.2 Vectorization with Triangulation
		2.3 Vectorization with Quadrangulation
	3 Algorithm
		3.1 Edge Extraction and Refinement
		3.2 Adaptive Mesh Generation
		3.3 Color Reconstruction
	4 Evaluation
	5 Applications
	6 Conclusions
	References
Temporal-Consistency-Aware Video Color Transfer
	1 Introduction
	2 Our Method
		2.1 Keyframe Color Transfer
		2.2 Video Interframe Color Transfer
		2.3 Abnormal Color Correction
		2.4 Luminance Smoothing
	3 Experimental Results and Discussions
	4 Conclusion and Future Work
	References
An Improved Advancing-front-Delaunay Method for Triangular Mesh Generation
	1 Introduction
	2 The Idea of Our Method
	3 Overall Method
	4 Node Generation
	5 Results and Discussion
	6 Conclusions and Future Research
	References
Robotics and Vision
Does Elderly Enjoy Playing Bingo with a Robot? A Case Study with the Humanoid Robot Nadine
	1 Introduction
	2 Related Work
	3 Experimental Setup
		3.1 Adaptation
		3.2 Participants
	4 Data Collection and Analysis
	5 Results
	6 Conclusion
	7 Discussion and Future Work
	References
Resilient Navigation Among Dynamic Agents with Hierarchical Reinforcement Learning
	1 Introduction
	2 Related Work
		2.1 Conventional Methods for Navigation
		2.2 Deep Reinforcement Learning Methods for Navigation
		2.3 Hierarchical Reinforcement Learning
	3 Approach
		3.1 Overview
		3.2 High-Level Module
		3.3 Low-Level Module
	4 Experiment
		4.1 Scene Design
		4.2 Perform Metrics
		4.3 Navigation Ability Comparison
		4.4 Resilience Comparison
		4.5 Ablation Experiment
	5 Conclusion
	References
Visual Analytics
MeshChain: Secure 3D Model and Intellectual Property management Powered by Blockchain Technology
	1 Introduction
	2 Related Work
		2.1 Decentralized Data Security
		2.2 Cooperative 3D Modeling
		2.3 Decentralized Version Control
		2.4 Decentralized Technologies in Computer Graphics
	3 Overview
	4 Cooperative Modeling
		4.1 Commit and Checkout
		4.2 Mesh Page
	5 Intellectual Property Management
		5.1 Mesh Incentive
		5.2 Data Authentication
	6 Results
		6.1 Decentralized Storage and Mesh Page
	7 Conclusion and Discussion
	References
Image Emotion Analysis Based on the Distance Relation of Emotion Categories via Deep Metric Learning
	1 Introduction
	2 Related Works
		2.1 Image Emotion Analysis
		2.2 Deep Metric Learning
	3 Method
		3.1 Pre-processing
		3.2 Sampling Mechanism
		3.3 Double-Weighted Mechanism
	4 Experiment
		4.1 Performance Analysis
		4.2 Ablation
		4.3 Similarity Decline Rule
	5 Conclusion
	References
How Much Do We Perceive Geometric Features, Personalities and Emotions in Avatars?
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Data Extraction
		3.2 Features Extraction
		3.3 Personality and Emotion Extraction
		3.4 Video Characteristics
	4 Results
		4.1 Geometric Features Perception
		4.2 Personality and Emotion Perceptions
	5 Final Considerations
	A  Appendix
		A.1  Visualization of Geometric Features
		A.2  Table of Scenes
		A.3  Questions
		A.4  Table of Results
	References
High-Dimensional Dataset Simplification by Laplace-Beltrami Operator
	1 Introduction
		1.1 Related Work
	2 High-Dimensional Dataset Simplification
		2.1 Spectrum of LBO and Feature Points
		2.2 Discrete Computation
		2.3 Dataset Simplification by Feature Point Detection
		2.4 Measurement of Fidelity of the Simplified Dataset
	3 Results and Discussion
		3.1 Simplification Results
		3.2 Quantitative Measurement of the Fidelity
		3.3 Comparisons
	4 Conclusion
	References
VR/AR
Characterizing Visual Acuity in the Use of Head Mounted Displays
	1 Introduction
	2 Related Work
	3 Methodology
		3.1 Hardware and Software Setup
		3.2 Study Protocol
		3.3 Snellen Test
		3.4 Pelli-Robson Contrast Sensitivity Test
		3.5 Central Glare
		3.6 Peripheral Glare
		3.7 Shooting Test
	4 Results
		4.1 Objective Results
		4.2 Subjective Results
		4.3 Discussion
	5 Conclusion
	References
Effects of Different Proximity-Based Feedback on Virtual Hand Pointing in Virtual Reality
	1 Introduction
	2 Related Work
		2.1 Depth Perception in Virtual Hand Pointing
		2.2 Feedback Cues for 3D Selections
	3 Experiment
		3.1 Participants
		3.2 Apparatus
		3.3 Material
		3.4 Experimental Design
		3.5 Task Procedure
		3.6 Metrics
	4 Result and Analysis
		4.1 Factor Analysis
		4.2 Subjective Evaluation
	5 Discussion
	6 Conclusion
	References
Virtual Scenes Construction Promotes Traditional Chinese Art Preservation
	1 Introduction
	2 Related Work
	3 Method
		3.1 Scene Primitive Model Generation Based on Semantics
		3.2 Scene Placement Algorithm Based on Priori Probability
	4 Implementation
		4.1 Realization of Scene Primitive Model
		4.2 Scene Placement
	5 Experiment and Discussion
	6 Conclusion
	References
A Preliminary Work: Mixed Reality-Integrated Computer-Aided Surgical Navigation System for Paranasal Sinus Surgery Using Microsoft HoloLens 2
	1 Introduction
		1.1 A Subsection Sample
	2 Method
		2.1 PC-CSN App
		2.2 PC-SS Extension
		2.3 MR App
		2.4 Experimental Setting
	3 Results
	4 Discussion
	5 Conclusion and Future Works
	References
Engage
Algorithms for Multi-conditioned Conic Fitting in Geometric Algebra for Conics
	1 Introduction
	2 Conic Fitting in GAC - Original Algorithm
		2.1 Implementation
	3 Conic Fitting in GAC - Additional Conditions
		3.1 Implementation
	4 Experimental Results and Applications
	5 Conclusion
	References
Special Affine Fourier Transform for Space-Time Algebra Signals
	1 Introduction
	2 Background
		2.1 Space-Time Algebra
		2.2 Space-Time Fourier Transform (SFT)
	3 Special Affine Space-Time Fourier Transform (SASFT)
		3.1 Defining the SASFT
		3.2 Properties of the SASFT
		3.3 Uncertainty Principle for the SASFT
	4 Conclusion
	References
On Explicit Formulas for Characteristic Polynomial Coefficients in Geometric Algebras
	1 Introduction
	2 Grade Projections and Operations of Conjugation in Geometric Algebras
	3 Characteristic Polynomials in Geometric Algebras
	4 The Cases n 4
	5 The Case n=5
	6 Conclusions
	References
Unified Expression Frame of Geodetic Stations Based on Conformal Geometric Algebra
	1 Introduction
	2 Basic Theory and Basic Idea
		2.1 Basic Theory
		2.2 Motion Operator
		2.3 Basic Idea
	3 Unified Expression Frame and ITRF Conversion
	4 Case Study
	5 Conclusion and Discussion
	References
Never `Drop the Ball' in the Operating Room: An Efficient Hand-Based VR HMD Controller Interpolation Algorithm, for Collaborative, Networked Virtual Environments
	1 Introduction
	2 State of the Art
	3 Room for Improvements
	4 Proposing New Approaches
		4.1 Proposed Method Based on Dual Quaternions
		4.2 Proposed Method Based on Multivectors
	5 Our Results
	6 Conclusions and Future Work
	References
The Rules of 4-Dimensional Perspective: How to Implement Lorentz Transformations in Relativistic Visualization
	1 Introduction
	2 The Rules of 4-Dimensional Perspective
	3 How to Implement Lorentz Transformations on a Computer
		3.1 Real Quaternions
		3.2 Complex Quaternions
		3.3 Lorentz Transformations
		3.4 4-Vectors
	4 Conclusion
	References
Author Index