Transfering Low-Frequency Features for Domain Adaptation

TL;DR

This paper introduces a low-frequency module (LFM) based on Gaussian low-pass filtering to improve unsupervised domain adaptation in computer vision by leveraging the domain-invariance of low-frequency features.

Contribution

The paper proposes a simple, hyperparameter-free low-frequency module that enhances domain adaptation by focusing on low-frequency features, compatible with existing methods.

Findings

LFM outperforms state-of-the-art methods in image classification.

LFM improves object detection across domains.

The approach is easy to implement and plug-and-play.

Abstract

Previous unsupervised domain adaptation methods did not handle the cross-domain problem from the perspective of frequency for computer vision. The images or feature maps of different domains can be decomposed into the low-frequency component and high-frequency component. This paper proposes the assumption that low-frequency information is more domain-invariant while the high-frequency information contains domain-related information. Hence, we introduce an approach, named low-frequency module (LFM), to extract domain-invariant feature representations. The LFM is constructed with the digital Gaussian low-pass filter. Our method is easy to implement and introduces no extra hyperparameter. We design two effective ways to utilize the LFM for domain adaptation, and our method is complementary to other existing methods and formulated as a plug-and-play unit that can be combined with these methods. Experimental results demonstrate that our LFM outperforms state-of-the-art methods for various computer vision tasks, including image classification and object detection.