[PDF] Deep Domain Adaptation For Visual Recognition With Unlabeled Data eBook

Author : Richa Singh
Publisher : Springer Nature
Page : 144 pages
File Size : 22,68 MB
Release : 2020-01-08
Category : Computers
ISBN : 3030306712

GET BOOK

This unique volume reviews the latest advances in domain adaptation in the training of machine learning algorithms for visual understanding, offering valuable insights from an international selection of experts in the field. The text presents a diverse selection of novel techniques, covering applications of object recognition, face recognition, and action and event recognition. Topics and features: reviews the domain adaptation-based machine learning algorithms available for visual understanding, and provides a deep metric learning approach; introduces a novel unsupervised method for image-to-image translation, and a video segment retrieval model that utilizes ensemble learning; proposes a unique way to determine which dataset is most useful in the base training, in order to improve the transferability of deep neural networks; describes a quantitative method for estimating the discrepancy between the source and target data to enhance image classification performance; presents a technique for multi-modal fusion that enhances facial action recognition, and a framework for intuition learning in domain adaptation; examines an original interpolation-based approach to address the issue of tracking model degradation in correlation filter-based methods. This authoritative work will serve as an invaluable reference for researchers and practitioners interested in machine learning-based visual recognition and understanding.

Author : Gabriela Csurka
Publisher : Springer Nature
Page : 182 pages
File Size : 50,19 MB
Release : 2022-06-06
Category : Computers
ISBN : 3031791754

GET BOOK

Solving problems with deep neural networks typically relies on massive amounts of labeled training data to achieve high performance. While in many situations huge volumes of unlabeled data can be and often are generated and available, the cost of acquiring data labels remains high. Transfer learning (TL), and in particular domain adaptation (DA), has emerged as an effective solution to overcome the burden of annotation, exploiting the unlabeled data available from the target domain together with labeled data or pre-trained models from similar, yet different source domains. The aim of this book is to provide an overview of such DA/TL methods applied to computer vision, a field whose popularity has increased significantly in the last few years. We set the stage by revisiting the theoretical background and some of the historical shallow methods before discussing and comparing different domain adaptation strategies that exploit deep architectures for visual recognition. We introduce the space of self-training-based methods that draw inspiration from the related fields of deep semi-supervised and self-supervised learning in solving the deep domain adaptation. Going beyond the classic domain adaptation problem, we then explore the rich space of problem settings that arise when applying domain adaptation in practice such as partial or open-set DA, where source and target data categories do not fully overlap, continuous DA where the target data comes as a stream, and so on. We next consider the least restrictive setting of domain generalization (DG), as an extreme case where neither labeled nor unlabeled target data are available during training. Finally, we close by considering the emerging area of learning-to-learn and how it can be applied to further improve existing approaches to cross domain learning problems such as DA and DG.

Author : Gabriela Csurka
Publisher : Springer
Page : 168 pages
File Size : 30,40 MB
Release : 2022-04-05
Category : Computers
ISBN : 9783031791802

GET BOOK

Solving problems with deep neural networks typically relies on massive amounts of labeled training data to achieve high performance. While in many situations huge volumes of unlabeled data can be and often are generated and available, the cost of acquiring data labels remains high. Transfer learning (TL), and in particular domain adaptation (DA), has emerged as an effective solution to overcome the burden of annotation, exploiting the unlabeled data available from the target domain together with labeled data or pre-trained models from similar, yet different source domains. The aim of this book is to provide an overview of such DA/TL methods applied to computer vision, a field whose popularity has increased significantly in the last few years. We set the stage by revisiting the theoretical background and some of the historical shallow methods before discussing and comparing different domain adaptation strategies that exploit deep architectures for visual recognition. We introduce the space of self-training-based methods that draw inspiration from the related fields of deep semi-supervised and self-supervised learning in solving the deep domain adaptation. Going beyond the classic domain adaptation problem, we then explore the rich space of problem settings that arise when applying domain adaptation in practice such as partial or open-set DA, where source and target data categories do not fully overlap, continuous DA where the target data comes as a stream, and so on. We next consider the least restrictive setting of domain generalization (DG), as an extreme case where neither labeled nor unlabeled target data are available during training. Finally, we close by considering the emerging area of learning-to-learn and how it can be applied to further improve existing approaches to cross domain learning problems such as DA and DG.

Author : Hemanth Venkateswara
Publisher : Springer Nature
Page : 256 pages
File Size : 45,48 MB
Release : 2020-08-18
Category : Computers
ISBN : 3030455297

GET BOOK

This book provides a survey of deep learning approaches to domain adaptation in computer vision. It gives the reader an overview of the state-of-the-art research in deep learning based domain adaptation. This book also discusses the various approaches to deep learning based domain adaptation in recent years. It outlines the importance of domain adaptation for the advancement of computer vision, consolidates the research in the area and provides the reader with promising directions for future research in domain adaptation. Divided into four parts, the first part of this book begins with an introduction to domain adaptation, which outlines the problem statement, the role of domain adaptation and the motivation for research in this area. It includes a chapter outlining pre-deep learning era domain adaptation techniques. The second part of this book highlights feature alignment based approaches to domain adaptation. The third part of this book outlines image alignment procedures for domain adaptation. The final section of this book presents novel directions for research in domain adaptation. This book targets researchers working in artificial intelligence, machine learning, deep learning and computer vision. Industry professionals and entrepreneurs seeking to adopt deep learning into their applications will also be interested in this book.

Author : Raghuraman Gopalan
Publisher :
Page : 93 pages
File Size : 40,74 MB
Release : 2015
Category : Computer vision
ISBN : 9781680830316

GET BOOK

Domain adaptation is an active, emerging research area that attempts to address the changes in data distribution across training and testing datasets. With the availability of a multitude of image acquisition sensors, variations due to illumination, and viewpoint among others, computer vision applications present a very natural test bed for evaluating domain adaptation methods. In this monograph, we provide a comprehensive overview of domain adaptation solutions for visual recognition problems. By starting with the problem description and illustrations, we discuss three adaptation scenarios namely, (i) unsupervised adaptation where the "source domain" training data is partially labeled and the "target domain" test data is unlabeled, (ii) semi-supervised adaptation where the target domain also has partial labels, and (iii) multi-domain heterogeneous adaptation which studies the previous two settings with the source and/or target having more than one domain, and accounts for cases where the features used to represent the data in each domain are different. For all these topics we discuss existing adaptation techniques in the literature, which are motivated by the principles of max-margin discriminative learning, manifold learning, sparse coding, as well as low-rank representations. These techniques have shown improved performance on a variety of applications such as object recognition, face recognition, activity analysis, concept classification, and person detection. We then conclude by analyzing the challenges posed by the realm of "big visual data", in terms of the generalization ability of adaptation algorithms to unconstrained data acquisition as well as issues related to their computational tractability, and draw parallels with the efforts from vision community on image transformation models, and invariant descriptors so as to facilitate improved understanding of vision problems under uncertainty.

Author : Gabriela Csurka
Publisher : Springer
Page : 338 pages
File Size : 17,16 MB
Release : 2017-09-10
Category : Computers
ISBN : 3319583476

GET BOOK

This comprehensive text/reference presents a broad review of diverse domain adaptation (DA) methods for machine learning, with a focus on solutions for visual applications. The book collects together solutions and perspectives proposed by an international selection of pre-eminent experts in the field, addressing not only classical image categorization, but also other computer vision tasks such as detection, segmentation and visual attributes. Topics and features: surveys the complete field of visual DA, including shallow methods designed for homogeneous and heterogeneous data as well as deep architectures; presents a positioning of the dataset bias in the CNN-based feature arena; proposes detailed analyses of popular shallow methods that addresses landmark data selection, kernel embedding, feature alignment, joint feature transformation and classifier adaptation, or the case of limited access to the source data; discusses more recent deep DA methods, including discrepancy-based adaptation networks and adversarial discriminative DA models; addresses domain adaptation problems beyond image categorization, such as a Fisher encoding adaptation for vehicle re-identification, semantic segmentation and detection trained on synthetic images, and domain generalization for semantic part detection; describes a multi-source domain generalization technique for visual attributes and a unifying framework for multi-domain and multi-task learning. This authoritative volume will be of great interest to a broad audience ranging from researchers and practitioners, to students involved in computer vision, pattern recognition and machine learning.

Author : Andrew Dudley (M.S.)
Publisher :
Page : 47 pages
File Size : 19,64 MB
Release : 2019
Category : Computer vision
ISBN :

GET BOOK

Endowing machines with the ability to understand digital images is a critical task for a host of high-impact applications, including pathology detection in radiographic imaging, autonomous vehicles, and assistive technology for the visually impaired. Computer vision systems rely on large corpora of annotated data in order to train task-specific visual recognition models. Despite significant advances made over the past decade, the fact remains collecting and annotating the data needed to successfully train a model is a prohibitively expensive endeavor. Moreover, these models are prone to rapid performance degradation when applied to data sampled from a different domain. Recent works in the development of deep adaptation networks seek to overcome these challenges by facilitating transfer learning between source and target domains. In parallel, the unification of dominant semi-supervised learning techniques has illustrated unprecedented potential for utilizing unlabeled data to train classification models in defiance of discouragingly meager sets of annotated data. In this thesis, a novel domain adaptation algorithm -- Domain Adaptive Fusion (DAF) -- is proposed, which encourages a domain-invariant linear relationship between the pixel-space of different domains and the prediction-space while being trained under a domain adversarial signal. The thoughtful combination of key components in unsupervised domain adaptation and semi-supervised learning enable DAF to effectively bridge the gap between source and target domains. Experiments performed on computer vision benchmark datasets for domain adaptation endorse the efficacy of this hybrid approach, outperforming all of the baseline architectures on most of the transfer tasks.

Author : Kevin Lin
Publisher :
Page : 131 pages
File Size : 28,61 MB
Release : 2020
Category :
ISBN :

GET BOOK

Supervised training with deep Convolutional Neural Networks (CNNs) have achieved great success in various visual recognition tasks. However, supervised training with deep CNNs requires large amount of well-annotated data. Data labeling, especially for large-scale image dataset, is very expensive. How to learn an effective model without the need of training data labeling has become an important problem for many applications. A promising solution is to create a learning protocol for the neural networks, so that the neural networks can learn to teach itself without manual labels. This technique is referred as the self-supervised learning, which has recently drawn an increasing attention for improving the learning performance. In this thesis, we first present our work on learning binary descriptors for fast image retrieval without manual labeling. We observe that images with the same category should have similar visual textures, and these similar textures are usually invariant to shift, scale and rotation. Thus, we could generate similar texture patch pairs automatically for training CNNs by shifting, scaling, and rotating image patches. Based on the observation, we design a training protocol for deep CNNs, which automatically generates pair-wise pseudo labels describing the similarity between the given two images. The proposed method performs more favorably than the baselines on different tasks including patch matching, image retrieval, and object recognition. In the second part of this thesis, we turn our focus to the task of human-centric analysis applications, and present our work on learning multi-person part segmentation without human labeling. Our proposed complementary learning technique learns a neural network model for multi-person part segmentation using a synthetic dataset and a real dataset. We observe that real and synthetic humans share a common skeleton structure. During learning, the proposed model extracts human skeletons which effectively bridges the synthetic and real domains. Without using human-annotated part segmentation labels, the resultant model works well on real world images. Our method outperforms the state-of-the-art approaches on multiple public datasets. Then, we discuss our work on accelerating multi-person pose estimation using a proposed concatenated pyramid network. We observe that each image may contain an unknown number of people that can occur at any scale or position. This makes fast multi-person pose estimation very challenging. Different from the earlier deep learning approaches that extract image features by using a series of convolutions, our proposed method extracts image features from each convolution layer in parallel, which better captures image features in different scales and improve the performance of human pose estimation. Our proposed method eliminates the need of multi-scale inference and multi-stage detection, and the proposed method is many times faster than the state-of-the-art approaches, while achieving better accuracy on the public datasets. Next, we present our work on 3D human mesh construction from a single image. We propose a novel approach to learn the human mesh representation without any ground truth mesh. This is made possible by introducing two new terms into the loss function of a graph convolutional neural network (Graph CNN). The first term is the Laplacian prior that acts as a regularizer on the mesh construction. The second term is the part segmentation loss that forces the projected region of the constructed mesh to match the part segmentation. Experimental results on multiple public datasets show that without using 3D ground truth meshes, the proposed approach outperforms the previous state-of-the-art approaches that require 3D ground truth meshes for training. Finally, we summarize our completed works and discuss the future research directions.