Computer vision -- ECCV 2020 : 16th European Conference, Glasgow, UK, August 23-28, 2020, Proceedings. Part III / Andrea Vedaldi, Horst Bischof, Thomas Brox, Jan-Michael Frahm (eds.).

Intro
Foreword
Preface
Organization
Contents
Part III
SIZER: A Dataset and Model for Parsing 3D Clothing and Learning Size Sensitive 3D Clothing
1 Introduction
2 Related Work
3 Dataset
3.1 SIZER Dataset: Scans
3.2 SIZER Dataset: SMPL and Garment Registrations
4 Method
4.1 Pose and Shape Prediction Network
4.2 ParserNet
4.3 SizerNet
4.4 Loss Functions
4.5 Implementation Details
5 Experiments and Results
5.1 Results of 3D Garment Parsing and Shape Under Clothing
5.2 Results of Garment Resizing
6 Conclusion
References

LIMP: Learning Latent Shape Representations with Metric Preservation Priors
1 Introduction
2 Related Work
3 Learning with Metric Priors
3.1 Losses
3.2 Continuous Interpretation
3.3 Implementation
4 Results
4.1 Data
4.2 Interpolation
4.3 Disentanglement
5 Conclusions
References
Unsupervised Sketch to Photo Synthesis
1 Introduction
2 Related Works
3 Unsupervised Two-Stage Sketch-to-Photo Synthesis
3.1 Shape Translation: Sketch S Grayscale G
3.2 Content Enrichment: Grayscale G Color I
4 Experiments and Applications

4.1 Experimental Setup and Evaluation Metrics
4.2 Sketch-Based Photo Synthesis Results
4.3 Ablation Study
4.4 Photo-to-Sketch Synthesis Results
4.5 Application: Unsupervised Sketch-Based Image Retrieval
References
A Simple Way to Make Neural Networks Robust Against Diverse Image Corruptions
1 Introduction
2 Related Work
3 Methods
3.1 Training with Gaussian Noise
3.2 Adversarial Noise
3.3 Evaluation on Corrupted Images
4 Results
5 Conclusions
References
SoftPoolNet: Shape Descriptor for Point Cloud Completion and Classification
1 Introduction

2 Related Work
3 Soft Pooling for Point Features
4 Regional Convolution
5 Network Architecture
6 Loss Functions
6.1 Learning Activations Through Regional Entropy
6.2 Reducing the Overlapping Regions
6.3 Preserving the Features from MLP
7 Experiments
7.1 Object Completion on ShapeNet
7.2 Car Completion on KITTI
7.3 Classification on ModelNet and PartNet
7.4 Ablation Study
8 Conclusion
References
Hierarchical Face Aging Through Disentangled Latent Characteristics
1 Introduction
2 Related Work
2.1 Face Aging
2.2 Variational Autoencoder

2.3 Age Estimation
3 Approach
3.1 Disentangled Variational Representations
3.2 Hierarchical Conditional Generator
3.3 Disentangled Adversarial Learning
3.4 Inference and Sampling
4 Experiments
4.1 Datasets and Settings
4.2 Qualitative Evaluation of DAAE
4.3 Quantitative Evaluation of DAAE
4.4 More Facial Age Analysis by DAAE
4.5 Ablation Study
5 Conclusion
References
Hybrid Models for Open Set Recognition
1 Introduction
2 Related Work
2.1 Open Set Recognition
2.2 Flow-Based Methods
2.3 Flow-Based Methods for Out-of-Distribution Detection