Advances in Visual Computing : 17th International Symposium, ISVC 2022, San Diego, CA, USA, October 3–5, 2022, Proceedings, Part II

The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Preface Organization Keynote Talks Towards Scaling Up GANs Sensible Machine Learning for Geometry Designing Augmented Reality for the Future of Work The Future of Visual Computing via Foundation Models (Banquet Keynote Talk) 3D Reconstruction: Leveraging Synthetic Data for Lightweight Reconstruction Human-AI Interaction in Visual Analytics: Designing for the “Two Black Boxes” Problem Contents – Part II Contents – Part I ST: Neuro-inspired Artificial Intelligence Brain Shape Correspondence Analysis Using Functional Maps 1 Introduction 2 Materials and Methods 2.1 Database 2.2 Methodology 3 Results 3.1 First Experiment 3.2 Second Experiment 3.3 Third Experiment 4 Conclusions References Biomimetic Oculomotor Control with Spiking Neural Networks 1 Introduction 2 Related Work 3 Eye Model and Neuromuscular Oculomotor Controller 4 Spiking Neurons 4.1 Encoding the Input Signals 4.2 Outputs 5 The SLiNet Model 5.1 Architecture 5.2 Training 6 Experiments 6.1 Eye Movements 6.2 Comparison to Human Eye Movements 7 Conclusions References Border Ownership, Category Selectivity and Beyond 1 Introduction 2 Implementation 2.1 Border-Ownership Coding Method 2.2 Category-Selective Coding Method 2.3 TcNet 3 Results 3.1 Datasets 3.2 Statistic Evaluation Criteria 4 Discussion 4.1 T-Junctions and Other ‘KEY’ Points 4.2 Global Context Awareness 4.3 Early Object Representation, ‘PRoto-Object’ 4.4 Relation to Biological Vision Systems 5 Summary References Sparse Kernel Transfer Learning 1 Introduction 2 Background 2.1 Background in Convolutional Neural Networks 2.2 Background in Sparse Coding 3 Methodology 3.1 Dictionary Learning 3.2 Initialization Techniques 3.3 Datasets 3.4 Kernel Transfer Learning 4 Experiments and Results 4.1 Comparison with Other Initialization Methods 4.2 Learning with Less Labels 4.3 Breast Cancer Detection 4.4 Intepretability and Complexity 5 Conclusion References Applications Photobombing Removal Benchmarking 1 Introduction 2 Related Work 2.1 Traditional Methods 2.2 Deep Learning-based Methods 3 Photobombing Removal Benchmark 3.1 Benchmarking Dataset 3.2 Benchmarking Methods 4 Experiments 4.1 Performance Metrics 4.2 Experimental Results 5 Conclusion and Future Works References Automatic Detection and Recognition of Products and Planogram Conformity Analysis in Real Time on Store Shelves 1 Introduction 1.1 Features for Detection of Retails Products 1.2 Detection of Single Product 2 Clustering by Products Famillies 2.1 Multi-object Detection with ASIFT 2.2 Distance Normalisation 2.3 DBSCAN: Products Famillies 2.4 Shelf Planogram Conformity Rate 3 Experiments 3.1 Database 3.2 Evaluation Metrics 4 Conclusion References Enhancing Privacy in Computer Vision Applications: An Emotion Preserving Approach to Obfuscate Faces 1 Introduction 2 Related Work 3 Approach 3.1 Face Detection 3.2 Face Selection 3.3 Face Reconstruction 3.4 Color Adaptation 3.5 Cloning 4 Validation 4.1 Experiment 4.2 Results 5 Conclusion and Future Work References House Price Prediction via Visual Cues and Estate Attributes 1 Introduction 2 Related Work 3 Proposed Work 3.1 Data Collection 3.2 Computational Model 4 Experiments 4.1 Evaluation Metrics 4.2 Experimental Results 4.3 Ablation Studies 5 Conclusion and Future Works References DRB-Net: Dilated Residual Block Network for Infrared Image Restoration 1 Introduction 2 Related Work 2.1 Non-learning Denoising Methods 2.2 Discriminative Learning Denoising Methods 2.3 Deep Learning for IR Imaging 3 Proposed Architecture 3.1 Why Dilated Convolution? 3.2 Residual Blocks 3.3 Architecture and Compared Methods 4 Dataset 4.1 Sample Preparation and Image Acquisition 4.2 Dataset Creation 4.3 Implementation 5 Experiments 5.1 DRB-Net Specification 5.2 Denoising of Synthetic Noisy Data 5.3 Generalization and Robustness Test 6 Conclusion and Future Work References Segmentation and Tracking Saliency Can Be All You Need in Contrastive Self-supervised Learning 1 Introduction 2 Motivation and Background 2.1 Related Work 2.2 Concrete Background 3 Implementation, Setup and Results 3.1 Setup and Datasets 3.2 Preliminary: Running SGD on NORCE-PV and MultiRes-PV Datasets 3.3 An Efficient Implementation 3.4 Using SGD as an Augmentation Policy in Contrastive SSL Algorithms 4 Discussion 5 Conclusions References GCEENet: A Global Context Enhancement and Exploitation for Medical Image Segmentation 1 Introduction 2 Related Work 2.1 Convolutional Neural Networks for Semantic Segmentation 2.2 Contextual Information Modeling 3 Proposed Architecture 3.1 Overview 3.2 Global Context Encoder Module 3.3 Local Distribution 3.4 Aggregator Module 3.5 Loss Function 4 Experiments and Discussion 4.1 Benchmark Datasets 4.2 Experiment Settings 5 Results and Discussion 5.1 Ablation Study 5.2 Comparison to Baseline Models 6 Conclusion References V2F: Real Time Video Segmentation with Apache Flink 1 Introduction 2 Related Work 3 Video2Flink Architecture 3.1 V2F Operators 4 Experiments 5 Conclusions and Future Work References Joint Discriminative and Metric Embedding Learning for Person Re-identification 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Classification Losses 3.2 Metric Learning Loss 3.3 Joint Classification and Metric Loss 3.4 Network Architecture 4 Experiments 4.1 Implementation Details 4.2 Comparison with State-of-the-Art Methods 4.3 Ablation Study 5 Conclusions References Transformer Networks for Future Person Localization in First-Person Videos 1 Introduction 2 Related Work 3 Proposed Method 3.1 Problem Overview 3.2 Input Overview 3.3 Implementation Details 4 Experiments 4.1 Evaluation Metrics and Baselines 4.2 Quantitative Results 4.3 Additional Analysis 4.4 Inference Time Analysis 5 Conclusion References Virtual Reality VR-SFT: Reproducing Swinging Flashlight Test in Virtual Reality to Detect Relative Afferent Pupillary Defect 1 Introduction 2 Literature Review 3 Methodology 3.1 Swinging Flashlight Test in Virtual Reality 3.2 VR Implementation and Experimental Software 3.3 RAPD Scoring 4 Dataset 5 Data Analysis and Results 6 Discussion and Future Work References A Quantitative Analysis of Redirected Walking in Virtual Reality Using Saccadic Eye Movements 1 Introduction 2 Methodology 2.1 Simulation and Hardware 2.2 Simulation Tasks and Data Collection 2.3 Eye Tracking 2.4 Questionnaire 2.5 Demographics 3 Results 4 Conclusion and Future Work References A DirectX-Based DICOM Viewer for Multi-user Surgical Planning in Augmented Reality 1 Introduction 2 Related Work 2.1 Holographic DICOM Viewer Prototypes 2.2 Interaction with 3D Objects 3 System Design Overview 4 Direct3D-Based DICOM Viewer Implementation 4.1 Smartphones as User Input Devices 4.2 Functionalities 4.3 Marker-Based 3D Object Placement 5 User Interactions 5.1 Virtual 2D Plane Touch 5.2 3D User Interaction 6 Experiments 7 Conclusions References Virtual-Reality Based Vestibular Ocular Motor Screening for Concussion Detection Using Machine-Learning 1 Introduction 2 Related Work 3 Methodology 3.1 Naive Bayes 3.2 Decision Tree 3.3 Random Forest 3.4 Support Vector Classifer 3.5 AdaBoost 3.6 Gaussian Process Classifier 3.7 Logistic Regression 3.8 Perceptron 3.9 Isolation Forest 3.10 One Class SVM 4 Experimental Analysis 4.1 Data Collection Using Virtual-Reality Headset 4.2 Data Splitting for Training and Testing 4.3 Qualitative Evaluation 4.4 Quantitative Evaluation 5 Conclusion References Posters GUILD - A Generator for Usable Images in Large-Scale Datasets 1 Introduction 2 Related Work 2.1 Manual Collection of Datasets 2.2 Synthetic Generation of Datasets 3 Implementation 3.1 Approach 3.2 Object Models 3.3 Environments 3.4 Label Generation 4 Evaluation 4.1 Evaluation Design 4.2 Evaluation Datasets 4.3 Accuracy 4.4 Generalizability 4.5 Variety 5 Conclusion and Future Work References Distributional Semantics of Line Charts for Trend Classification 1 Introduction 2 Dataset 3 Related Work 3.1 Information Graphic Description Generation 3.2 Prototype Learning 3.3 Bag of Words for Computer Vision 3.4 Distributional Semantics 4 Architecture and Methodology 4.1 Forming the Vocabulary 4.2 Line Chart Embeddings 4.3 Classification 5 Implementation 6 Experiments and Results 6.1 Classification Task 6.2 Results 7 Discussion 8 Conclusion References Deep Learning Hyperparameter Optimization for Breast Mass Detection in Mammograms 1 Introduction 2 Background and Motivation 2.1 End-to-End Pipeline 2.2 Genetic Algorithm 2.3 Binary Tournament Selection 2.4 Simulated Binary Crossover (SBX) 2.5 Polynomial Mutation 3 Approach 3.1 Pre-processing Images 3.2 Building Spark Cluster 3.3 GA-E2E Algorithm 3.4 Evaluation 4 Experimental Results 4.1 Experimental Setup 4.2 Running GA-E2E 4.3 Training Evaluation 4.4 Analysis 5 Experimental Issues 6 Conclusion and Future Work 6.1 Conclusion 6.2 Future Work References Analysis of Deep Learning-Based Image Steganalysis Methods Under Different Steganographic Algorithms 1 Introduction 2 Related Work 3 Proposed Method 3.1 CNN Architecture 3.2 Filter Banks 4 Experiments Setup 4.1 Dataset Description 4.2 Experiments Description 5 Results 5.1 Experiment 1 Results 5.2 Experiment 2 Results 5.3 Experiment 3 Results 5.4 Experiment 4 Results 6 Conclusion References Emotion Recognition in Video Streams Using Intramodal and Intermodal Attention Mechanisms 1 Introduction 2 Related Work 3 Proposed Approach 3.1 The Visual Self-attention (VSA) 3.2 The Audio Self-Attention (ASA) 3.3 The Cross-Modal Fusion (CMF) 4 Experimental Evaluation 5 Conclusions and Perspectives References Driver State Detection from In-Car Camera Images 1 Introduction 2 Related Work 3 Proposed Method 3.1 Training Data 3.2 Architecture of Anomaly Detector 4 Experiments 5 Conclusion References Transferability Limitations for Covid 3D Localization Using SARS-CoV-2 Segmentation Models in 4D CT Images 1 Introduction 2 Related Work 2.1 Our Contribution 3 Experimental Setup 3.1 Datasets Description 3.2 Experimental Results 4 Dataset Limitations 5 3D Representations 6 Conclusions References Deep Architecture Based Spalling Severity Detection System Using Encoder-Decoder Networks 1 Introduction 1.1 Related Works 1.2 Contributions 2 Research Methodology 2.1 UNet 2.2 SegNet 2.3 Data Prepossessing and Augmentation 2.4 Proposed Deep Architecture 3 Result 3.1 Data Processing and Experimental Setup 3.2 Quantitative Analysis 3.3 Qualitative Analysis 4 Conclusion References Overview on Machine Vision Based Surface Defect Detection and Quality Classification in the Leather Manufacturing Process 1 Introduction 2 Related Work 2.1 Supervised Learning Methods (Deep Learning Methods) 2.2 Semi-supervised and Unsupervised Methods 2.3 Image Datasets 3 Challenges and Constraints 3.1 Challenges and Constraints in Dataset Acquisition and Pre-processing 3.2 Challenges and Constraints in Anomaly Detection Method 4 Recommendations 4.1 Recommendations for Dataset Acquisition and Pre-processing 4.2 Recommendations for Feature Extraction 4.3 Recommendations for Anomaly Detection 4.4 Recommendations for Evaluation Methods and Techniques 5 Conclusion References Efficient Shadow Removal and Enhancement of Light Texts in Scanned Documents 1 Introduction 2 Proposed Method 2.1 Compute Document Foreground Mask 2.2 Compute Document Shadow Map 2.3 Remove Shadows 2.4 Text Enhancement 3 Experimental Results 3.1 Datasets 3.2 Results and Discussion 4 Conclusion References A Game Theoretical Vulnerability Analysis of Adversarial Attack 1 Introduction 2 Methodology 2.1 CNN Model and Loss Function 2.2 FGSM Attack 2.3 One Pixel Attack 2.4 Optimization: Differential Evolution(DE) 3 Experimental Analysis 3.1 Dataset 3.2 Hyper-parameter Tuning 3.3 Training Procedure 4 Analysis: Game Theory 4.1 Game Formulation 4.2 A Characterization of Threshold for Game Strategy 4.3 Utility Value and Simulation Results on CAPTCHA Data 5 Conclusion References 2D Fingertip Localization on Depth Videos Using Paired Video-to-Video Translation 1 Introduction 2 Related Work 2.1 Domain Transfer Learning for Hand Keypoint Localization 2.2 Optical Flow 3 Methodology 3.1 Mathematics 3.2 Implementation Details on Video to Video Translation 3.3 Color Segmentation in HSV Color Space 4 Experiments 4.1 Data Preparation 4.2 Evaluation Metrics 5 Qualitative and Quantitative Results 5.1 Quantitative Comparison 5.2 Qualitative Result 6 Conclusion References Difference-in-level Detection from RGB-D Images 1 Introduction 2 Related Work 2.1 Difference-in-level Detection 2.2 Edge Detection from RGB-D Images with Deep Learning 3 Method 3.1 Difference-in-level Detection for Making the Dataset 3.2 Difference-in-level Detection with Convolutional Neural Network 4 Experiments 4.1 Difference-in-level Detection 4.2 Qualitative Evaluation on Various Inputs 4.3 Ablation Study on Inputs 4.4 Comparison with Existing Methods 5 Conclusion References A Contour Extraction Method for Garment Recognition Based on Improved Segmentation and Gabor Filter 1 Introduction 2 Contour Extraction Methods 2.1 Moore-Neighbor Tracking Algorithm 2.2 Gabor Filter Based Contour Extraction Method 2.3 Reference Region Based Extraction Method 3 Dataset for Performance Verification 4 Results 4.1 Contours Extracted with Moore-Neighbor Tracking Algorithm 4.2 Contours Extracted with Gabor Filter Based Contour Extraction Method 4.3 Contours Extracted with the Proposed Self-adaptive Method 4.4 Quantitative Analysis 5 Conclusion References A Robust Deep Transfer Learning Model for Accurate Speech Emotion Classification 1 Introduction 2 Literature Review and Related Work 3 Methods and Techniques 3.1 DCNN for Speech Emotion Classification 3.2 Extraction of Features 3.3 Principal Component Analysis (PCA) 3.4 Multilayer Perceptron (MLP) Classifier 3.5 Random Forest Classifier (RF) 4 Experimental Results and Analysis 4.1 Datasets 4.2 Results and Discussion 4.3 Performance Evaluation 5 Conclusion References Analysis of Smooth Pursuit Assessment in Virtual Reality and Concussion Detection Using BiLSTM 1 Introduction 2 Related Work 3 Methodology 3.1 Data Collection 3.2 Network Description 3.3 Proposed Metric 4 Experiments 4.1 Dataset and Preprocessing 4.2 Performance Metric 4.3 Hyper-parameter Tuning 5 Results & Discussion 6 Conclusion and Future Work References Author Index The two volume set LNCS 8887 and 8888 constitutes the refereed proceedings of the 10th International Symposium on Visual Computing, ISVC 2014, held in Las Vegas, NV, USA. The 74 revised full papers and 55 poster papers presented together with 39 special track papers were carefully reviewed and selected from more than 280 submissions. The papers are organized in topical sections: Part I (LNCS 8887) comprises computational bioimaging, computer graphics; motion, tracking, feature extraction and matching, segmentation, visualization, mapping, modeling and surface reconstruction, unmanned autonomous systems, medical imaging, tracking for human activity monitoring, intelligent transportation systems, visual perception and robotic systems. Part II (LNCS 8888) comprises topics such as computational bioimaging , recognition, computer vision, applications, face processing and recognition, virtual reality, and the poster sessions. This two-volume set of LNCS 13598 and 13599 constitutes the refereed proceedings of the 17th International Symposium on Visual Computing, ISVC 2022, which was held in October 2022. The 61 papers presented in these volumes were carefully reviewed and selected from 110 submissions. They are organized in the following topical sections: Part I: ​deep learning I; visualization; object detection and recognition; deep learning II; video analysis and event recognition; computer graphics; ST: biomedical imaging techniques for cancer detection, diagnosis and management. Part II: ​ST: neuro-inspired artificia intelligence; applications; segmentation and tracking; virtual reality; poster.