Deep Visual Domain Adaptation: A Survey

By Mei Wang, Weihong Deng. 2018

2,330 citations recorded by the source corpus.

Deep domain adaption has emerged as a new learning technique to address the lack of massive amounts of labeled data. Compared to conventional methods, which learn shared feature subspaces or reuse important source instances with shallow representations, deep domain adaption methods leverage deep networks to learn more transferable representations by embedding domain adaptation in the pipeline of deep learning. There have been comprehensive surveys for shallow domain adaption, but few timely reviews the emerging deep learning based methods. In this paper, we provide a comprehensive survey of deep domain adaptation methods for computer vision applications with four major contributions. First, we present a taxonomy of different deep domain adaption scenarios according to the properties of data that define how two domains are diverged. Second, we summarize deep domain adaption approaches into several categories based on training loss, and analyze and compare briefly the state-of-the-art methods under these categories. Third, we overview the computer vision applications that go beyond image classification, such as face recognition, semantic segmentation and object detection. Fourth, some potential deficiencies of current methods and several future directions are highlighted.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Radiology, Nuclear Medicine and Imaging

Citations in the corpus

A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark (2019, background): “Improving sample efﬁciency has been approached from many angles: few-shot learning (Fei-Fei et al., 2006), transfer learning (Pan & Yang, 2009), domain adaptation (Wang & Deng, 2018), and representation learning (Bengio et al., 2013).”
Learning Robust Global Representations by Penalizing Local Predictive Power (2019, methodology): “For a general primer, we refer the reader to the following reviews (Weiss et al., 2016; Csurka, 2017; Wang and Deng, 2018).”
A Comprehensive Survey on Transfer Learning (2019, neutral): “Besides, a number of surveys focus on specific application scenarios including activity recognition [25], visual categorization [26], collaborative recommendation [27], computer vision [24], and sentiment analysis [28].”
Recent Advances in Open Set Recognition: A Survey (2018, neutral): “As most of them followed the evaluation protocol8 defined in [89] and did not provide source codes, similar to [3], [147], we here only compare with their published results.”
PixelVAE++: Improved PixelVAE with Discrete Prior (2019, background): “Many machine learning tasks such as domain adaptation [12, 13, 14, 15] and few-shot learning [16, 17, 18] rely on learned representations of observations.”
An Information-Theoretic Approach to Transferability in Task Transfer Learning (2019, background): “…where knowledge transfer is achieved by making representations learned from one input domain work on a different input domain, e.g. adapt models for RGB images to infrared images [21]; and task transfer learning , where knowledge is transferred between different tasks on the same input domain [22].”
Deep Face Recognition: A Survey (2018, background): “Re-collecting massive labeled data to train a new model from scratch or re-collecting some unlabeled data to perform domain adaptation [166] or others are effective methods.”
Generalized Out-of-Distribution Detection: A Survey (2021, neutral): “Domain Adaptation/Generalization Domain Adaptation (DA) (Wang & Deng, 2018) and Domain Generalization (DG) (Zhou et al., 2021b) also follow “open-world” assumption.”
Virtual to Real Adaptation of Pedestrian Detectors (2020, background): “An exhaustive survey about deep learning DA techniques is provided in [28].”
Attending to Discriminative Certainty for Domain Adaptation (2019, methodology): “To address this issue, some methods have been proposed for adapting domains [7, 50].”
The Emerging Trends of Multi-Label Learning (2020, background): “For instance, there might be out-of-distribution detection [134], domain shift [135] and other problems arise inMLCproblems.”
Transfer learning: a friendly introduction (2022, background): “However, in the last stage, among the tested image, recovered best-suiting image and the Markov model transfers, the cross-domain sets with a best-matching image to test [58].”
The Visual Task Adaptation Benchmark (2019, background): “However, it would be interesting to research alternative strategies from domain adaptation (Wang & Deng, 2018), residual adapters (Rebuffi et al., 2017), transfusion (Raghu et al., 2019), and other fields.”
Stochastic Classifiers for Unsupervised Domain Adaptation (2020, background): “There have been a large spectrum of UDA methods developed in the literature [36, 56].”
A Review of Domain Adaptation without Target Labels (2019, background): “We refer to [24], [25], [26] for extensive reviews.”
A Review of Single-Source Deep Unsupervised Visual Domain Adaptation (2020, background): “Similar to [41–43], we focus on deep domain adaptation, but use a different taxonomy that provides different insights.”