Geometry and vision : first International Symposium, ISGV 2021, Auckland, New Zealand, January 28-29, 2021, Revised Selected Papers / / edited by Minh Nguyen, Wei Qi Yan, and Harvey Ho

• Pubblicazione: Cham, Switzerland : , : Springer, , [2021]; ©2021
• Descrizione fisica: 1 online resource (407 pages) : illustrations
• Disciplina: 006.37
• Soggetto topico: Computer vision
• Persona (resp. second.): HoHarvey <1983->; YanWei Qi; NguyễnMinh
• Note generali: Includes index.
• Nota di contenuto: Intro -- Preface -- Organization -- Contents -- A New Noise Generating Method Based on Gaussian Sampling for Privacy Preservation -- 1 Introduction -- 2 Related Work -- 2.1 Gaussian Noise Generating -- 2.2 Whittle's Noise Estimator -- 2.3 The Method Based on Fourier Transform -- 2.4 Distributed SGD for Differential Privacy -- 3 Our Methods -- 3.1 Contribution -- 3.2 The Process of Our Methods -- 3.3 Noise Variant in Stochastic Gradient Descent -- 3.4 Gaussian Distribution for Subsampling -- 4 Evaluations -- 4.1 Our Experiments for Comparing Learning Rates -- 4.2 Experiments for Gradient Clipping and Noise Levels -- 5 Conclusions -- References -- Traffic-Sign Recognition Using Deep Learning -- 1 Introduction -- 2 Literature Review -- 3 Methodology -- 3.1 Data Collection -- 3.2 Research Design for Training Faster R-CNN -- 3.3 Research Design for Training Faster YOLOv5 -- 4 Results -- 4.1 Experiment Results of Faster R-CNN -- 4.2 Experiment Results of YOLOv5 -- 5 Analysis -- 6 Conclusion and Future Work -- References -- Tree Leaves Detection Based on Deep Learning -- 1 Introduction -- 1.1 Background and Motivation -- 1.2 Contribution -- 2 Literature Review -- 3 Methodology -- 3.1 Working Principle and Structure Analysis of YOLO -- 3.2 Analysis of the Working Principle of Faster R-CNN -- 3.3 Environmental Deployment -- 3.4 Data Set Preparation -- 3.5 Evaluation Methods -- 4 Analysis and Discussions -- 4.1 Comparison of Object Detection Results -- 4.2 Comparative Analysis of the Two Proposed Models -- 4.3 Discussions -- 5 Conclusion and Future Work -- References -- Deep Learning in Medical Applications: Lesion Segmentation in Skin Cancer Images Using Modified and Improved Encoder-Decoder Architecture -- 1 Introduction -- 2 Related Study -- 3 Materials and Methods -- 3.1 Encoder-Decoder Framework -- 3.2 Network Architecture Details. ; 4 Simulations and Results Discussion -- 4.1 Dataset Preparation -- 4.2 Quality Metrics -- 4.3 Comparison with Other State-of-the-Art Methods -- 5 Discussion and Conclusion -- References -- Apple Ripeness Identification Using Deep Learning -- 1 Introduction -- 2 Related Work -- 3 Our Approaches -- 4 Our Experiments -- 5 Conclusion -- References -- A Hand-Held Sensor System for Exploration and Thermal Mapping of Volcanic Fumarole Fields -- 1 Introduction -- 2 Related Work -- 3 Data Acquisition -- 3.1 Sensor System and Registration -- 3.2 Datasets -- 3.3 Thermal Sensing Quality -- 4 Methods -- 4.1 Localization -- 4.2 3D Reconstruction - Direct -- 4.3 3D Reconstruction - Indirect -- 5 Results -- 6 Discussion -- 7 Conclusion -- References -- Traffic Sign Recognition Using Guided Image Filtering -- 1 Introduction -- 2 Literature Review -- 3 Network Design -- 3.1 Guided Image Filtering -- 3.2 Improved Faster R-CNN -- 3.3 Improved YOLOv5 -- 4 Results -- 4.1 Improved Faster R-CNN -- 4.2 Improved YOLOv5 -- 4.3 Comparsions of YOLOv5 and Faster R-CNN -- 4.4 Guided Image Filtering -- 5 Conclusion -- References -- Towards a Generic Bicubic Hermite Mesh Template for Cow Udders -- 1 Introduction -- 2 Methods -- 2.1 Data Cloud of Cow Udders -- 2.2 Bicubic Hermite Mesh -- 2.3 Coherent Point Drifting -- 3 Results -- 3.1 Morphing the CH Mesh of the Udder -- 3.2 Geometric Modelling for the Teat -- 4 Discussion -- 5 Conclusion -- References -- Sign Language Recognition from Digital Videos Using Deep Learning Methods -- 1 Introduction -- 2 Related Work -- 3 Our Method -- 4 Experimental Results -- 5 Conclusion and Future Work -- References -- New Zealand Shellfish Detection, Recognition and Counting: A Deep Learning Approach on Mobile Devices -- 1 Introduction and Backgrounds -- 2 Conceptualisation of Implementation Method -- 2.1 Overall System Design. ; 2.2 Dataset Preparation -- 2.3 Detection Model Design -- 2.4 Web Application Deployment Design -- 3 Experiment -- 3.1 Data Pre-processing -- 3.2 Model Implementation -- 4 Analysis and Discussion -- 4.1 Model Comparison -- 4.2 Results Demonstration -- 5 Conclusion and Future Work -- References -- Coverless Video Steganography Based on Inter Frame Combination -- 1 Introduction -- 2 The Proposed Method -- 2.1 Generating Hash Sequence -- 2.2 Mapping Rule -- 2.3 Information Hiding -- 2.4 Information Extraction -- 3 Experiment Results and Analysis -- 3.1 Capacity -- 3.2 Robustness -- 3.3 Security Analysis -- 4 Conclusion -- References -- Character Photo Selection for Mobile Platform -- 1 Introduction -- 2 Related Work -- 2.1 Feature Extraction for Person -- 2.2 Photo Selection -- 3 Method -- 3.1 Proposed Framework -- 3.2 Elimination Stage -- 3.3 Selection Stage -- 4 Experiments and Analysis -- 4.1 Dataset -- 4.2 Features Importance Ranking -- 4.3 Experimental Comparison and Analysis -- 5 Conclusion -- References -- Close Euclidean Shortest Path Crossing an Ordered 3D Skew Segment Sequence -- 1 Introduction -- 2 Preliminaries -- 3 Euclidean Shortest Path Crossing a Sequence of 3D Skew Segments -- 4 Conclusion -- References -- A Lane Line Detection Algorithm Based on Convolutional Neural Network -- 1 Introduction -- 2 Method -- 2.1 Algorithm Framework -- 2.2 The Encoder -- 2.3 The Decoder -- 3 The Implementation Process -- 3.1 Remove the Full Connection Layer -- 3.2 Increase the Cavity Convolution -- 3.3 Instance Segmentation -- 3.4 Afterprocessing -- 4 Experimental Results and Analysis -- 4.1 Lane Line Detection Results -- 5 Conclusion -- References -- Segment- and Arc-Based Vectorizations by Multi-scale/Irregular Tangential Covering -- 1 Introduction -- 2 Reconstructions into Maximal Primitives -- 2.1 Multi-scale Noise Detection. ; 2.2 Irregular Isothetic Cyclic Representation -- 2.3 Recognition of Line Segments and Circular Arcs -- 3 Adapted Tangential Covering -- 3.1 The minDSS Algorithm -- 3.2 Adaptation of minDSS -- 4 Experimental Results -- 4.1 Global Overview of the Method -- 4.2 Visual Inspection of Results with Synthetic Images -- 4.3 Visual Inspection of Results with a Real Image -- 5 Conclusion and Future Works -- References -- Algorithms for Computing Topological Invariants in Digital Spaces -- 1 Introduction -- 2 Background Concepts of Digital Spaces -- 3 Hole Counting Algorithms in 2D -- 3.1 The Simple Formula for the Number of Holes in S -- 3.2 Algorithms for Hole Counting -- 4 Algorithms and Implementations for the Genus of Digital Surfaces in 3D -- 4.1 Practical Algorithms and Implementations -- 4.2 Implementations and Data Samples -- 5 Remarks on Programming -- 6 Summary and Discussion -- References -- Discrete Linear Geometry on Non-square Grid -- 1 Introduction -- 2 Hexagonal Grid System on a Plane -- 3 Reconstruction of Euclidean Line -- 4 Polygonalisation from Hexels -- 5 Numerical Examples -- 6 Conclusions -- References -- Electric Scooter and Its Rider Detection Framework Based on Deep Learning for Supporting Scooter-Related Injury Emergency Services -- 1 Introduction -- 1.1 Motivation -- 1.2 Proposed Idea -- 2 Background -- 2.1 Traditional Object Detection Algorithms -- 2.2 Deep Learning-Based Object Detection Algorithms -- 3 Design and Implementation -- 3.1 E-Scooter and Its Rider Detection Framework -- 3.2 E-Scooter and Its Rider Detection Model Training and Deploying -- 3.3 Fall Detection Implementation -- 4 Results and Evaluations -- 4.1 Training Model Process -- 4.2 Evaluation Model Process -- 5 Conclusion and Future Work -- References -- Tracking Livestock Using a Fully Connected Network and Kalman Filter -- 1 Introduction -- 2 Related Work. ; 3 Methodology -- 3.1 Object Tracker -- 3.2 Data Association -- 3.3 New and Old Tracks -- 4 Experimental Evaluation -- 4.1 Parameter Selection -- 4.2 Metrics -- 4.3 Results -- 5 Conclusion -- References -- A Comparison of Approaches for Synchronizing Events in Video Streams Using Audio -- 1 Introduction -- 2 Background -- 2.1 The Mel Spectrogram -- 2.2 Template Matching with Cross-Correlation -- 3 Data Collection -- 3.1 Database of Videos -- 3.2 Manual Annotation for Training -- 4 Methods -- 4.1 Deep Learning Model -- 4.2 Template Matching Model -- 4.3 Re-creation of Trial and Evaluation Metric -- 5 Results and Discussions -- 5.1 Deep Learning Outcome -- 5.2 Template Matching Outcome -- 5.3 Reconstruction of Trial Times from Detected Events -- 6 Conclusion -- References -- Union-Retire: A New Paradigm for Single-Pass Connected Component Analysis -- 1 Introduction -- 1.1 Union-Find Algorithms -- 1.2 Contributions -- 2 Prior Work -- 3 Union-Retire Algorithm -- 3.1 Data Structures -- 3.2 Algorithm Description -- 3.3 Example -- 4 Analysis -- 4.1 Correctness -- 4.2 Validation -- 4.3 Memory Requirements -- 4.4 Computational Complexity -- 5 Summary and Conclusion -- References -- Improving Object Detection in Real-World Traffic Scenes -- 1 Introduction -- 2 Literature Review -- 3 Methodology -- 3.1 Precision and Recall -- 3.2 Score System -- 3.3 SSD Based Models -- 4 Results and Discussion -- 4.1 Model 1: Suitable CT Identification Using Default SSD -- 4.2 Model 2: Suitable Resolution Using Default SSD and CT = 0.3 -- 4.3 Model 3: Preprocessing Comparison on SSD Models with CT = 0.3 and Resolution = [800 × 600] Pixels -- 5 Conclusions -- References -- Comparison of Red versus Blue Laser Light for Accurate 3D Measurement of Highly Specular Surfaces in Ambient Lighting Conditions -- 1 Introduction -- 2 Commercial Solutions -- 3 Methodology. ; 4 Experiments and Results.
• Titolo autorizzato: Geometry and vision
• ISBN: 3-030-72073-X
• Formato: Materiale a stampa
• Livello bibliografico: Monografia
• Lingua di pubblicazione: Inglese
• Record Nr.: 9910484794803321
• Lo trovi qui: Univ. Federico II
• Serie: Communications in Computer and Information Science