Image and Video Technology: 10th Pacific-Rim Symposium, PSIVT 2022, Virtual Event, November 12–14, 2022, Proceedings (Lecture Notes in Computer Science, 13763)

This book constitutes the conference proceedings of the 10th Pacific Rim Symposium on Image and Video Technology, PSIVT 2022, held in Bintan Island, Indonesia, in November 2022. A total of 15 papers were carefully reviewed and selected from 18 submissions. The main conference focuses on theoretical advances or practical implementations in image and video technology. Preface Organization Contents Waste Classification from Digital Images Using ConvNeXt 1 Introduction 2 Related Work 3 Our Method 4 Result Analysis 4.1 Our Dataset 4.2 Evaluation Methods 4.3 Result Analysis 4.4 Ablation Experiments 5 Conclusion References A Federated Learning Approach for Text Classification Using NLP 1 Introduction 2 Related Work 3 Proposed Methodology 3.1 Dataset Description 3.2 Data Pre-processing 3.3 GRU 3.4 Fed Averaging 3.5 Bi-LSTM 3.6 CNN 4 Experiments and Results 5 Conclusion References A Method for Face Image Inpainting Based on Autoencoder and Generative Adversarial Network 1 Introduction 2 Literature Review 3 Methodology 3.1 Generative Network 3.2 Discriminator Network 3.3 Algorithms 4 Result Analysis 5 Conclusion References Traffic Sign Recognition from Digital Images by Using Deep Learning 1 Introduction 2 Literature Review 2.1 External Factors Affecting Traffic Sign Recognition 2.2 Related Algorithms 3 Our Method 3.1 Dark Channel Prior Defogging Algorithm 3.2 Guided Image Filtering 3.3 YOLOv5 Model for Traffic Signs Recognition 4 Our Results 4.1 Data Sources and Data Collection 4.2 Comparison and Analysis of Two Defogging Model 4.3 TSR Experiments 4.4 Result Comparisons 4.5 Analysis and Discussions 5 Conclusion References Youtube Engagement Analytics via Deep Multimodal Fusion Model 1 Introduction 2 Related Works 3 Our Model 3.1 Feature Extraction 3.2 Multi-modal 4 Experiments 4.1 Dataset and Evaluation Metrics 4.2 Experimental Settings 4.3 Results 5 Discussion 6 Conclusion References Dynamic Point Cloud Compression with Cross-Sectional Approach 1 Introduction 1.1 Introducing V-PCC Method 1.2 Limitations of the Current Standard Dynamic Point Cloud Compression Technology 1.3 Contributions and Benefits of the Proposed Cross-Sectional Approach 2 Proposed Method 2.1 Five Key Priorities for Cross-Sectional Segmentation 2.2 Detailed Cross-Sectional Process 2.3 Possibility for Further Segmentation to Achieve Even Greater Quality 3 Experimental Results 3.1 Visual Comparisons of the Methods 3.2 Objective Evaluations Comparing Both Methods 4 Conclusion References Event-Based Visual Sensing for Human Motion Detection and Classification at Various Distances 1 Introduction 1.1 Contribution 1.2 Related Work 2 Methods 2.1 Dataset 2.2 Network Architecture and Training Method 2.3 Detection Algorithm Using a Density-Based Algorithm 3 Experimental Results 3.1 Classification Accuracy 3.2 Detection on Multiple Targets 4 Discussion 5 Conclusion References On Low-Resolution Face Re-identification with High-Resolution-Mapping 1 Introduction 2 Related Works 3 Proposed Protocol for HR-Mapping 4 Methods for HR-Mapping 5 Proposed Dataset Protocol 6 Experiments 7 Conclusions References On Skin Lesion Recognition Using Deep Learning: 50 Ways to Choose Your Model 1 Introduction 2 Related Works 3 Evaluated Methods 4 Experimental Results 4.1 Dataset 4.2 Results 4.3 Discussion 4.4 Implementation 5 Conclusions References Improving Automated Baggage Inspection Using Simulated X-ray Images of 3D Models 1 Introduction 2 Proposed Method 2.1 X-ray Image Simulation 2.2 Colorization 2.3 Superimposition Preparation 2.4 Superimposition 2.5 Object Detection 3 Experimental Setup 3.1 X-ray Image Simulation 3.2 Colorization 3.3 Superimposition 3.4 Object Detection 3.5 Implementation 4 Results and Discussion 5 Conclusions References A Wasserstein GAN for Joint Learning of Inpainting and Spatial Optimisation 1 Introduction 1.1 Our Contribution 1.2 Related Work 1.3 Organisation of the Paper 2 Inpainting with Wasserstein GANs 3 Learning Masks with Wasserstein GANs 3.1 Learning Binary Masks 3.2 Joint Learning of Inpainting Operator and Masks 4 Experiments 4.1 Experimental Methodology 4.2 Comparison Against NetM 4.3 Comparison Against Probabilistic Methods 5 Conclusion and Future Work References Rapid On-Site Weed Identification with Machine Learning 1 Introduction 2 Related Work 3 Proposed Approach 3.1 Weeds Selection and Image Data Acquisition 3.2 ID Model Development Unit 4 Experimental Results and Discussion 4.1 Experiment Setup 4.2 Implementation and Results 4.3 Discussion 5 Conclusions References Remote Tiny Weeds Detection 1 Introduction 2 Related Work 3 Methodology 3.1 Study Sites 3.2 Image Data Acquisition 3.3 Automatic Detection of Organge Hawkweed 4 Proposed Deep Models 5 Results and Analysis 5.1 Training and Testing 5.2 Performance Evaluation 6 Discussion and Conclusion References Combining Multi-vision Embedding in Contextual Attention for Vietnamese Visual Question Answering 1 Introduction 2 Related Works 3 Methodology 3.1 Multi-vision Embedding for Image Understanding 3.2 Question Understanding 3.3 Multi-branch Contextual Attention 3.4 Answer Prediction 3.5 Loss Function 4 Experiment 4.1 Dataset and Evaluation Metric 4.2 Experiment Settings 4.3 Comparison with Other Methods 4.4 Ablation Study 5 Conclusion References Depth Estimation of Traffic Scenes from Image Sequence Using Deep Learning 1 Introduction 2 Literature Review 3 Our Methods 4 Experimental Results 5 Conclusion References Author Index