Computer Vision and Graphics: Proceedings of the International Conference on Computer Vision and Graphics ICCVG 2022

✍ Scribed by Leszek J. Chmielewski, Arkadiusz Orłowski

• Publisher: Springer
• Year: 2023
• Tongue: English
• Leaves: 264
• Series: Lecture Notes in Networks and Systems, 598
• Category: Library

✦ Synopsis

This book contains 17 papers presented at the conference devoted to cutting-edge technologies and concepts related to image processing. A broad collection of problems including man–machine interfaces, comparison of quantum and conventional computing in deep learning, medical image processing, image segmentation, face recognition, outdoor scene analysis, image rendering and colorization, map generation, traffic analysis, hardware acceleration, data association, and visual cryptography is investigated.

Research on these issues is important, among others due to that large amounts of video data are collected continually. They can be easily stored, but their analysis is still a challenge.

The book is primarily intended for researchers and practitioners in image analysis and generation, as well as for students in the fields related to computer science. However, any reader interested in the subject matter of the book will find some chapters interesting and valuable.

✦ Table of Contents

Organization
Preface
Contents
Computers and Humans
On Multi-stream Classification of Two Person Interactions in Video with Skeleton-Based Features
1 Introduction
2 Related Work
3 Approach
3.1 Structure
3.2 Key Frame Selection
3.3 Skeleton Detection
3.4 Skeleton Tracking and Correcting
3.5 Feature Extraction
4 Ensemble of Weak'' Classifiers 5 Triple Stream LSTM 6 Results 6.1 Dataset 6.2 Ensemble of Experts 6.3 Multi-stream LSTM 6.4 Ablation Study 7 Conclusion References On Formal Models of Interactions Between Detectors and Trackers in Crowd Analysis Tasks 1 Introduction 2 Formal Models 2.1 Basic Definitions 2.2 Matrix-Based Data Associations Between Monitoring Events 2.3 Estimates of Association Matrices 2.4 From Association Matrices to Labeling 3 Discussion References Digital Wah-Wah Guitar Effect Controlled by Mouth Movements 1 Introduction 2 The Principle of the Wah-Wah Effect 3 Conception 4 Initial Experiments 5 Research Environment 6 Results 7 Summary References Traffic and Driving Traffic Sign Classification Using Deep and Quantum Neural Networks 1 Introduction 2 Quantum Neural Networks 3 Previous Work 4 QNN for Traffic Sign Recognition 5 Conclusion References Real Time Intersection Traffic Queue Length Estimation System 1 Introduction 2 Methodology 2.1 Module Which Creates Configuration for Selected Camera (Configuration Mode) 2.2 Module Which Estimates the Queue Length Based on Configuration 3 Experiments 4 Conclusion and Further Plans References Image Processing Carotid Artery Wall Segmentation in Ultrasound Image Sequences Using a Deep Convolutional Neural Network 1 Introduction 2 Datasets 3 Segmentation Method 3.1 Detection of the Far Wall 3.2 Segmentation of the IMC 4 Results 5 Discussion 6 Compliance with Ethical Standards Information References Novel Co-SIFT Detector for Scanned Images Differentiation 1 Introduction 2 Related Work 2.1 Analysis of 8 Methods Utilizing Color for SIFT or SURF 3 Contactless Scanning 4 Transformation to Assignment Problem 5 Co-SIFT 5.1 Preprocessing 5.2 Interesting Point Detection 6 Experiments and Results 7 Conclusions 8 Future Work References PointPillars Backbone Type Selection for Fast and Accurate LiDAR Object Detection 1 Introduction 2 Related Work 2.1 DCNN Methods for 3D Object Detection on a LiDAR Point Cloud 2.2 Methods for Real-Time Object Detection on Images 3 Comparison of Backbone Types for the PointPillars Network 4 Results 5 Conclusions References Fuzzy Approach to Object-Detection-Based Image Retrieval 1 Introduction 2 Related Works 3 Fuzzy Modeling Object Mutual Positions 4 Querying Image Database 4.1 Single Query with Complete Relation 4.2 Combined Queries 4.3 Queries with Incomplete Relations 4.4 Complex Queries and Parsing 5 Tests 6 Conclusions References Adaptive Binarization of Metal Nameplate Images Using the Pixel Voting Approach 1 Introduction 2 A Brief Review of Image Binarization Methods 2.1 Global and Adaptive Thresholding Algorithms 2.2 Recent Proposals 3 Proposed Approach 3.1 Pixel Voting 3.2 Dataset ofIndustrial'' Images
4 Results
5 Conclusions
References
Influence of Step Parameterisation on the Results of the Reidentification Pipeline
1 Introduction
2 Related Work
3 Materials and Methods
3.1 Pipeline Under Study
3.2 Metrics Used
4 Results
5 Conclusions
References
On the Influence of Image Features on the Performance of Deep Learning Models in Human-Object Interaction Detection
1 Introduction
2 Related Works
3 Baseline Approach
3.1 Object Detection and Feature Extraction
3.2 Human and Object Streams
3.3 Relation Stream and Final Score Calculation
4 Experiments
4.1 Experimental Setup
4.2 Human Stream
4.3 Human and Object Streams
4.4 Object Category and Distance Information
4.5 Summary of Results
5 Conclusions
References
Computer Graphics
Fast Triangle Strip Generation and Tunneling for Different Cost Metrics
1 Introduction
2 Background
3 Related Work
4 Cost Metrics
5 Implementation
5.1 Overview
5.2 Stripify Stage
5.3 Tunneling Stage
5.4 Novel Loop Detection Algorithm
6 Benchmarking
6.1 Comparison of Implementations
6.2 Results
6.3 Extended Tunneling
7 Conclusion and Future Work
8 Appendix
References
A Deep Multi-Layer Perceptron Model for Automatic Colourisation of Digital Grayscale Images
1 Introduction
2 Related Works
3 Methods
4 Results
5 Conclusion and Future Works
References
An Algorithm for Automatic Creation of Ground Level Maps in Two-Dimensional Top-Down Digital Role Playing Games
1 Introduction
2 A Brief Description of the General Approach
3 The Algorithm for the Creation of the Ground Level Map
4 The Two Rules Algorithm
4.1 The Double Edges Rule
4.2 The Middle Rule
5 Results
6 Conclusions and Future Plans
References
Hardware and Cryptography
Energy Efficient Hardware Acceleration of Neural Networks with Power-of-Two Quantisation
1 Introduction
2 Power-of-Two Quantisation
3 Related Work
4 Hardware Design
4.1 Benchmark Results
4.2 Going Lower on Power—Pruning
5 Conclusion
References
Error Analysis and Graphical Evidence of Randomness in Two Methods of Color Visual Cryptography
1 Introduction
2 Two Methods of Coding
2.1 Description
2.2 Algorithms
2.3 Errors
2.4 Example Images
3 Probability of Errors
3.1 Theoretical Background
3.2 Experiment
3.3 Results and Discussion
4 Randomness Tests
4.1 Presenting the Histograms of p-Values
4.2 Presenting the Results of Tests with Subtests
4.3 Results and Discussion
5 Conclusions
References
Appendix Author Index
Index