Computer Vision – ECCV 2022 Workshops : Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part VI / edited by Leonid Karlinsky, Tomer Michaeli, Ko Nishino.

Format:

Book

Contributor:

Karlinsky, Leonid, editor.

Michaeli, Tomer, editor.

Nishino, Ko, editor.

Series:

Lecture Notes in Computer Science, 1611-3349 ; 13806

Language:

English

Subjects (All):

Computer vision.

Pattern recognition systems.

Machine learning.

Computer networks.

Social sciences--Data processing.

Social sciences.

User interfaces (Computer systems).

Human-computer interaction.

Computer Vision.

Automated Pattern Recognition.

Machine Learning.

Computer Communication Networks.

Computer Application in Social and Behavioral Sciences.

User Interfaces and Human Computer Interaction.

Local Subjects:

Computer Vision.

Automated Pattern Recognition.

Machine Learning.

Computer Communication Networks.

Computer Application in Social and Behavioral Sciences.

User Interfaces and Human Computer Interaction.

Physical Description:

1 online resource (805 pages)

Edition:

1st ed. 2023.

Place of Publication:

Cham : Springer Nature Switzerland : Imprint: Springer, 2023.

Summary:

The 8-volume set, comprising the LNCS books 13801 until 13809, constitutes the refereed proceedings of 38 out of the 60 workshops held at the 17th European Conference on Computer Vision, ECCV 2022. The conference took place in Tel Aviv, Israel, during October 23-27, 2022; the workshops were held hybrid or online. The 367 full papers included in this volume set were carefully reviewed and selected for inclusion in the ECCV 2022 workshop proceedings. They were organized in individual parts as follows: Part I: W01 - AI for Space; W02 - Vision for Art; W03 - Adversarial Robustness in the Real World; W04 - Autonomous Vehicle Vision Part II: W05 - Learning With Limited and Imperfect Data; W06 - Advances in Image Manipulation; Part III: W07 - Medical Computer Vision; W08 - Computer Vision for Metaverse; W09 - Self-Supervised Learning: What Is Next?; Part IV: W10 - Self-Supervised Learning for Next-Generation Industry-LevelAutonomous Driving; W11 - ISIC Skin Image Analysis; W12 - Cross-Modal Human-Robot Interaction; W13 - Text in Everything; W14 - BioImage Computing; W15 - Visual Object-Oriented Learning Meets Interaction: Discovery, Representations, and Applications; W16 - AI for Creative Video Editing and Understanding; W17 - Visual Inductive Priors for Data-Efficient Deep Learning; W18 - Mobile Intelligent Photography and Imaging; Part V: W19 - People Analysis: From Face, Body and Fashion to 3D Virtual Avatars; W20 - Safe Artificial Intelligence for Automated Driving; W21 - Real-World Surveillance: Applications and Challenges; W22 - Affective Behavior Analysis In-the-Wild; Part VI: W23 - Visual Perception for Navigation in Human Environments: The JackRabbot Human Body Pose Dataset and Benchmark; W24 - Distributed Smart Cameras; W25 - Causality in Vision; W26 - In-Vehicle Sensing and Monitorization; W27 - Assistive Computer Vision and Robotics; W28 - Computational Aspectsof Deep Learning; Part VII: W29 - Computer Vision for Civil and Infrastructure Engineering; W30 - AI-Enabled Medical Image Analysis: Digital Pathology and Radiology/COVID19; W31 - Compositional and Multimodal Perception; Part VIII: W32 - Uncertainty Quantification for Computer Vision; W33 - Recovering 6D Object Pose; W34 - Drawings and Abstract Imagery: Representation and Analysis; W35 - Sign Language Understanding; W36 - A Challenge for Out-of-Distribution Generalization in Computer Vision; W37 - Vision With Biased or Scarce Data; W38 - Visual Object Tracking Challenge. .

Contents:

Intro

Foreword

Preface

Organization

Contents - Part VI

W22 - Competition on Affective Behavior Analysis In-the-Wild

Geometric Pose Affordance: Monocular 3D Human Pose Estimation with Scene Constraints

1 Introduction

2 Related Work

3 Geometric Pose Affordance Dataset (GPA)

4 Geometry-Aware Pose Estimation

4.1 Pose Estimation Baseline Model

4.2 Geometric Consistency Loss and Encoding

4.3 Overall Training

5 Experiments

5.1 Baselines

5.2 Effectiveness of Geometric Affordance

6 Discussion and Conclusion

References

Affective Behaviour Analysis Using Pretrained Model with Facial Prior

3 Proposed Method

3.1 Formulation

3.2 EMMA

3.3 Masked CoTEX

4 Experiment

4.1 Benchmarks and Evaluation Metrics

4.2 Training Details

4.3 Recognition Results

4.4 Ablation Study

5 Conclusions

Facial Affect Recognition Using Semi-supervised Learning with Adaptive Threshold

2 Method

2.1 MFAR: Multi-task Facial Affect Recognition

2.2 SS-MFAR: Semi-supervised Multi-task Facial Affect Recognition

2.3 Problem Formulation

2.4 Adaptive Threshold

2.5 Supervision Loss

2.6 Unsupervised and Consistency Loss

2.7 Overall Loss

3 Dataset and Implementation Details

3.1 Dataset

3.2 Implementation Details

3.3 Evaluation Metrics

4 Results and Discussion

4.1 Performance on Validation Set

4.2 Ablation Studies

4.3 Performance on Test Set

MT-EmotiEffNet for Multi-task Human Affective Behavior Analysis and Learning from Synthetic Data

2 Proposed Approach

2.1 Multi-task Learning Challenge

2.2 Learning from Synthetic Data Challenge

3 Experimental Study.

3.1 FER for Static Images

3.2 Multi-task-Learning Challenge

3.3 Learning from Synthetic Data Challenge

4 Conclusions

Ensemble of Multi-task Learning Networks for Facial Expression Recognition In-the-Wild with Learning from Synthetic Data

2.1 Data Pre-processing

2.2 Model Architecture

3 Experiments and Results

3.1 Training Setup

3.2 Performance Evaluation

4 Conclusion

PERI: Part Aware Emotion Recognition in the Wild

3 Method

3.1 MediaPipe Holistic Model

3.2 The Emotic Model

3.3 Part Aware Spatial Image

3.4 Context Infusion Blocks

4 Experiments

4.1 Experiment Setup

4.2 Quantitative Results

4.3 Qualitative Results

5 Conclusion

Facial Expression Recognition with Mid-level Representation Enhancement and Graph Embedded Uncertainty Suppressing

3.1 Overview

3.2 Mid-level Representation Enhancement

3.3 Graph Embedded Uncertainty Suppressing

3.4 Latent Distribution Mining and Pairwise Uncertainty Estimation

3.5 Mixed Loss Functions

3.6 Prediction Merging and Correcting

4.1 Dataset

4.2 Settings

4.3 Results

Deep Semantic Manipulation of Facial Videos

3.1 AI Tool

3.2 Person-Specific Dataset of Facial Expressions

3.3 Expression Decoder Network

3.4 Synthesis of Photorealistic Manipulated videos

4 Experimental Results

4.1 Qualitative Evaluation

4.2 Quantitative Evaluation

4.3 User Studies

5 Conclusion and Future Work

BYEL: Bootstrap Your Emotion Latent

2 Related Works

2.1 Self-supervised Representation Learning.

2.2 Human Emotion Analysis

3 Problem Description

4 Method

4.1 Bootstrap Your Emotion Latent

4.2 Transfer-Learning

4.3 Loss

5.1 Dataset

5.2 Settings

5.3 Metric

5.4 Results

6 Conclusions

Affective Behavior Analysis Using Action Unit Relation Graph and Multi-task Cross Attention

3.1 AU Relation Graph

3.2 FER and VA Estimation Heads

3.3 Loss Function

4.2 Evaluation Metrics

4.3 Experiment Setting

4.4 Result

Multi-Task Learning Framework for Emotion Recognition In-the-Wild

3.1 Features

3.2 Temporal Encoder

3.3 Single Task Loss Function

3.4 Multi-Task Learning Framework

4.2 Experiment Setup

4.3 Overall Results on the Validation Set

4.4 Model Ensemble

4.5 Results on the Test Set

ABAW: Learning from Synthetic Data &amp

Multi-task Learning Challenges

2 Competition Corpora

3 Evaluation Metrics for Each Challenge

3.1 Multi-task Learning Challenge

3.2 Learning from Synthetic Data Challenge

4 Baseline Networks and Performance

4.1 Multi-task Learning Challenge

4.2 Learning from Synthetic Data Challenge

Two-Aspect Information Interaction Model for ABAW4 Multi-task Challenge

2 Methodology

2.1 ROI Feature Extraction

2.2 Interaction Module

2.3 Temporal Smoothing and Classifier Module

2.4 Losses

3 Experiments

3.1 Datasets

3.2 Training Details

3.3 Evaluation Metric

3.4 Baseline Model

4 Results

4.1 Comparison with Baseline Models.

4.2 Different Temporal Smoothing Methods

4.3 Different Backbones

4.4 Different Semantic Information

4.5 Detail Results

Facial Expression Recognition In-the-Wild with Deep Pre-trained Models

W23 - Visual Perception for Navigation in Human Environments: The JackRabbot Human Body Pose Dataset and Benchmark

Robustness of Embodied Point Navigation Agents

3.1 Habitat Environment

3.2 PointNav Task

3.3 Metrics

3.4 Visual Corruptions

3.5 Dynamics Corruptions

4 Experimental Setup

4.1 Agents

4.2 Dataset

4.3 Configurations

5 Results

6 Conclusion and Limitations

W24 - Distributed Smart Cameras

CounTr: An End-to-End Transformer Approach for Crowd Counting and Density Estimation

2.1 Crowd Counting and Density Estimation

2.2 Transformers in Vision

3 Methodology

3.1 Preliminaries on Transformers for Crowd Counting

3.2 The CounTr Model

4.1 Implementation Details and Datasets

4.2 Results and Discussion

4.3 Ablation Study

4.4 Visualization on Density Maps

On the Design of Privacy-Aware Cameras: A Study on Deep Neural Networks

3 Experimental Study

3.1 Experimental Setup

3.2 Semantic Segmentation Sensitivity to Color and Defocus Blur

3.3 LPDR Sensitivity to Defocus Blur

RelMobNet: End-to-End Relative Camera Pose Estimation Using a Robust Two-Stage Training

1 Introduction.

2 Related Work

2.1 Feature Correspondence Methods

2.2 End-to-End Methods

3.1 Dataset Image Pairs

3.2 Architecture Details

4 Experiments and Analysis

4.1 Baseline Comparison

4.2 Two-Stage vs One-Stage Approach

4.3 Secondary Data Collection and Evaluation

Cross-Camera View-Overlap Recognition

3 Decentralising the Recognition of View Overlaps

3.1 Camera Messaging

3.2 Sharing Query Views

3.3 View-Overlap Recognition

4 Validation

4.1 Strategies Under Comparison

4.2 Implementation Details

4.3 Dataset and View-Overlap Annotation

4.4 Performance Measures

4.5 Parameter Setting

4.6 Discussion

Activity Monitoring Made Easier by Smart 360-degree Cameras

3 Equirectangular Images and Their Geometry

3.1 From Equirectangular Image to Ground-Plane Coordinates

4 Activity Monitoring in 360-degree Camera Network

4.1 Correspondences Between the Camera Reference Systems

4.2 Detecting Activities Within Areas of Interests

4.3 Detecting Areas of Interest in the Scene

5 Implementation Details

6 Experiments

6.1 Finding Correspondences Between Camera Reference Systems

6.2 Activity Detection Within RoI

6.3 Detecting Areas of Interest

7 Conclusions and Future Work

Seeing Objects in Dark with Continual Contrastive Learning

2 Proposed Method

3 Related Work and Experiments

3.1 Ablation Studies, and Detailed Hyperparameter Sensitivity Analysis of Our Method

3.2 Comparison with State-of-the-Art

Towards Energy-Efficient Hyperspectral Image Processing Inside Camera Pixels

2.1 Energy-Efficient On-Device Vision.

2.2 HSI Camera Operations.

Notes:

Includes bibliographical references and index.

Other Format:

Print version: Karlinsky, Leonid Computer Vision - ECCV 2022 Workshops

ISBN:

9783031250750

3031250753