Domain Adaptation for Semantic Segmentation via Patch-Wise Contrastive Learning

TL;DR

This paper presents a contrastive learning-based method for domain adaptation in semantic segmentation, effectively aligning features across domains and outperforming existing methods, especially with limited annotations.

Contribution

Introduces a contrastive learning approach for domain adaptation in semantic segmentation that outperforms adversarial methods and reduces annotation costs.

Findings

Outperforms state-of-the-art methods on adaptive segmentation tasks.

Effective with limited target domain annotations.

Extensible to weakly-supervised domain adaptation.

Abstract

We introduce a novel approach to unsupervised and semi-supervised domain adaptation for semantic segmentation. Unlike many earlier methods that rely on adversarial learning for feature alignment, we leverage contrastive learning to bridge the domain gap by aligning the features of structurally similar label patches across domains. As a result, the networks are easier to train and deliver better performance. Our approach consistently outperforms state-of-the-art unsupervised and semi-supervised methods on two challenging domain adaptive segmentation tasks, particularly with a small number of target domain annotations. It can also be naturally extended to weakly-supervised domain adaptation, where only a minor drop in accuracy can save up to 75% of annotation cost.