OVANet: One-vs-All Network for Universal Domain Adaptation

ICCV 2021  ·  Kuniaki Saito, Kate Saenko ·

Universal Domain Adaptation (UNDA) aims to handle both domain-shift and category-shift between two datasets, where the main challenge is to transfer knowledge while rejecting unknown classes which are absent in the labeled source data but present in the unlabeled target data. Existing methods manually set a threshold to reject unknown samples based on validation or a pre-defined ratio of unknown samples, but this strategy is not practical. In this paper, we propose a method to learn the threshold using source samples and to adapt it to the target domain. Our idea is that a minimum inter-class distance in the source domain should be a good threshold to decide between known or unknown in the target. To learn the inter-and intra-class distance, we propose to train a one-vs-all classifier for each class using labeled source data. Then, we adapt the open-set classifier to the target domain by minimizing class entropy. The resulting framework is the simplest of all baselines of UNDA and is insensitive to the value of a hyper-parameter yet outperforms baselines with a large margin.

PDF Abstract ICCV 2021 PDF ICCV 2021 Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Universal Domain Adaptation DomainNet OVANet H-Score 50.7 # 6
Source-free no # 1
Universal Domain Adaptation Office-31 OVANet H-score 86.5 # 9
Source-Free no # 1
Universal Domain Adaptation Office-Home OVANet H-Score 71.8 # 8
Source-free no # 1
Universal Domain Adaptation VisDA2017 OVANet H-score 53.1 # 9
Source-free no # 1


No methods listed for this paper. Add relevant methods here