Learning the Roots of Visual Domain Shift
By Tatiana Tommasi, Martina Lanzi, P. Russo, Barbara Caputo. 2016
38 citations recorded by the source corpus.
In this paper we focus on the spatial nature of visual domain shift, attempting to learn where domain adaptation originates in each given image of the source and target set. We borrow concepts and techniques from the CNN visualization literature, and learn domainness maps able to localize the degree of domain specificity in images. We derive from these maps features related to different domainness levels, and we show that by considering them as a preprocessing step for a domain adaptation algorithm, the final classification performance is strongly improved. Combined with the whole image representation, these features provide state of the art results on the Office dataset.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition
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
- Barbara Caputo
- Saeid Motiian
- Gianfranco Doretto
- Q. Jones
- Seyed Mehdi Iranmanesh
- Piotr Koniusz
- Elisa Ricci
- Fatih Murat Porikli
In Computer Vision and Pattern Recognition
- Xiaogang Wang
- Luc Van Gool
- Barbara Caputo
- Ming-Hsuan Yang
- Dacheng Tao
- Trevor Darrell
- Tao A. Xiang
- Gianfranco Doretto
Citations in the corpus
- Few-Shot Adversarial Domain Adaptation (2017, background): “Some try to find a function which can map from the source domain to the target domain [47, 28, 19, 16, 11, 55, 52].”
- Unified Deep Supervised Domain Adaptation and Generalization (2017, background): “The first one includes those that try to find a mapping between source and target distributions [53, 34, 27, 26, 19, 57].”
- AutoDIAL: Automatic Domain Alignment Layers (2017, methodology): “As baselines we consider Geodesic Flow Kernel (GFK) [11], Subspace Alignment (SA) [7]), CORAL [26], Transfer Component Analysis (TCA) [24], Simultaneous Deep Transfer (SDT) [30], and the recent method in [28].”
- Domain Adaptation by Mixture of Alignments of Second-or Higher-Order Scatter Tensors (2016, methodology): “In [38], the authors utilize socalled “domainness maps” which capture locally the degree of domain specificity.”
- Unlimited Road-scene Synthetic Annotation (URSA) Dataset (2018, background): “This causes domain shift [13]—a difference in image generating distributions between the natural and synthetic images.”