Linked e-resources
Details
Table of Contents
Intro
Foreword
Preface
Organization
Contents
Part XI
MessyTable: Instance Association in Multiple Camera Views
1 Introduction
2 Related Work
3 MessyTable Dataset
3.1 Variations in View Angles
3.2 Variations in Scenes
3.3 Data Annotation
4 Baselines
4.1 Appearance Information
4.2 Surrounding Information
4.3 Geometric Methods
5 Experiments
5.1 Evaluation Metrics
5.2 Benchmarking Baselines on MessyTable
5.3 Effects of View Variation and Scene Complexity
5.4 MessyTable as a Benchmark and a Training Source
6 Discussion
References
A Unified Framework for Shot Type Classification Based on Subject Centric Lens
1 Introduction
2 Related Work
3 MovieShots Dataset
3.1 Shot Categories
3.2 Dataset Statistics
3.3 Annotation Procedure
4 SGNet: Subject Guidance Network
4.1 Subject Map Guidance Classification
4.2 Subject Map Generation
5 Experiments
5.1 Experiments Setup
5.2 Overall Results
5.3 Ablation Studies
5.4 Qualitative Results
6 Application
7 Conclusion
References
BSL-1K: Scaling Up Co-articulated Sign Language Recognition Using Mouthing Cues
1 Introduction
2 Related Work
3 Learning Sign Recognition with Automatic Labels
3.1 Finding Probable Signing Windows in Public Broadcast Footage
3.2 Precise Sign Localisation Through Visual Keyword Spotting
3.3 BSL-1K Dataset Construction and Validation
4 Models and Implementation Details
4.1 Visual Keyword Spotting Model
4.2 Sign Recognition Model
4.3 Video Pose Distillation
5 Experiments
5.1 Ablations for the Sign Recognition Model
5.2 Benchmarking Sign Recognition and Sign Spotting
5.3 Comparison with the State of the Art on ASL Benchmarks
6 Conclusion
References
HTML: A Parametric Hand Texture Model for 3D Hand Reconstruction and Personalization
1 Introduction
2 Related Work
3 Textured Parametric Hand Model
3.1 Data Acquisition
3.2 Data Canonicalization
3.3 Texture Model Creation
4 Applications
4.1 3D Hand Personalization from a Single Image
4.2 Self-supervised Photometric Loss
5 Experiments
5.1 Texture Model Evaluation
5.2 Application Results: 3D Hand Personalization
5.3 Application Results: Photometric Neural Network Loss
6 Limitations and Discussion
7 Conclusion
References
CycAs: Self-supervised Cycle Association for Learning Re-identifiable Descriptions
1 Introduction
2 Related Work
3 Proposed Cycle Association Task
3.1 Overview
3.2 Association Between Symmetric Pairs
3.3 Relaxation on Asymmetric Pairs
3.4 Adapt Softmax Temperature to Varying Sizes
3.5 Two-Stage Training
4 Experiment
4.1 Experiment with Standard Datasets
4.2 Ablation Study
4.3 Experiment Using Self-collected Videos as Training Data
5 Conclusion
References
Open-Edit: Open-Domain Image Manipulation with Open-Vocabulary Instructions
1 Introduction
Foreword
Preface
Organization
Contents
Part XI
MessyTable: Instance Association in Multiple Camera Views
1 Introduction
2 Related Work
3 MessyTable Dataset
3.1 Variations in View Angles
3.2 Variations in Scenes
3.3 Data Annotation
4 Baselines
4.1 Appearance Information
4.2 Surrounding Information
4.3 Geometric Methods
5 Experiments
5.1 Evaluation Metrics
5.2 Benchmarking Baselines on MessyTable
5.3 Effects of View Variation and Scene Complexity
5.4 MessyTable as a Benchmark and a Training Source
6 Discussion
References
A Unified Framework for Shot Type Classification Based on Subject Centric Lens
1 Introduction
2 Related Work
3 MovieShots Dataset
3.1 Shot Categories
3.2 Dataset Statistics
3.3 Annotation Procedure
4 SGNet: Subject Guidance Network
4.1 Subject Map Guidance Classification
4.2 Subject Map Generation
5 Experiments
5.1 Experiments Setup
5.2 Overall Results
5.3 Ablation Studies
5.4 Qualitative Results
6 Application
7 Conclusion
References
BSL-1K: Scaling Up Co-articulated Sign Language Recognition Using Mouthing Cues
1 Introduction
2 Related Work
3 Learning Sign Recognition with Automatic Labels
3.1 Finding Probable Signing Windows in Public Broadcast Footage
3.2 Precise Sign Localisation Through Visual Keyword Spotting
3.3 BSL-1K Dataset Construction and Validation
4 Models and Implementation Details
4.1 Visual Keyword Spotting Model
4.2 Sign Recognition Model
4.3 Video Pose Distillation
5 Experiments
5.1 Ablations for the Sign Recognition Model
5.2 Benchmarking Sign Recognition and Sign Spotting
5.3 Comparison with the State of the Art on ASL Benchmarks
6 Conclusion
References
HTML: A Parametric Hand Texture Model for 3D Hand Reconstruction and Personalization
1 Introduction
2 Related Work
3 Textured Parametric Hand Model
3.1 Data Acquisition
3.2 Data Canonicalization
3.3 Texture Model Creation
4 Applications
4.1 3D Hand Personalization from a Single Image
4.2 Self-supervised Photometric Loss
5 Experiments
5.1 Texture Model Evaluation
5.2 Application Results: 3D Hand Personalization
5.3 Application Results: Photometric Neural Network Loss
6 Limitations and Discussion
7 Conclusion
References
CycAs: Self-supervised Cycle Association for Learning Re-identifiable Descriptions
1 Introduction
2 Related Work
3 Proposed Cycle Association Task
3.1 Overview
3.2 Association Between Symmetric Pairs
3.3 Relaxation on Asymmetric Pairs
3.4 Adapt Softmax Temperature to Varying Sizes
3.5 Two-Stage Training
4 Experiment
4.1 Experiment with Standard Datasets
4.2 Ablation Study
4.3 Experiment Using Self-collected Videos as Training Data
5 Conclusion
References
Open-Edit: Open-Domain Image Manipulation with Open-Vocabulary Instructions
1 Introduction