Abstract:The semantic segmentation of high-resolution (HR) remote sensing images (RSIs) has been extensively researched in various applications. However, segmentation networks are prone to significant performance degradation on unlabeled data due to domain shift, such as data distribution shifts arising from distinct geographic locations. To address this issue, we propose a multilevel-guided curriculum domain adaptation (MuGCDA) approach for joint samplewise, categorywise, and pixelwise tasks, which facilitates the final fine-grained segmentation task by guiding the target domain to acquire samplewise and categorywise domain-robust properties. Concretely, at the sample level, we formulate a sample spatial relationship consistency guidance (SSCG) loss that guides the target domain to acquire similar sample spatial relationship properties to the source domain. At the category level, we propose a category layout structure consistency guidance (CLCG) module that guides the target domain to acquire consistent layout properties. At the pixel level, we design an adaptive hierarchical pseudolabel weight setting (AHPWS) method with a self-training (ST) paradigm to reduce the effect of label noise while improving the quality of the generated pseudolabels. Furthermore, to improve the stability of the training process, we use a momentum network (MN) as the teacher network to obtain the property knowledge and pseudolabels, and then guide the whole domain transfer process of the segmentation network, which acts as the student network. Extensive comparison and ablation experiments are conducted in several cross-space and cross-spectral scenes, and the results show that our method achieves significant performance improvements in cross-domain scenes for HR RSIs.

Dual Path Learning for Domain Adaptation of Semantic Segmentation.

When Masked Image Modeling Meets Source-free Unsupervised Domain Adaptation: Dual-Level Masked Network for Semantic Segmentation

Bidirectional Learning for Domain Adaptation of Semantic Segmentation

Domain Adaptive Semantic Segmentation with Self-Supervised Depth Estimation

Focus on Your Target: A Dual Teacher-Student Framework for Domain-adaptive Semantic Segmentation

Semi-supervised Domain Adaptation for Semantic Segmentation

Target-targeted Domain Adaptation for Unsupervised Semantic Segmentation

Domain Adaptation for Remote Sensing Image Semantic Segmentation: An Integrated Approach of Contrastive Learning and Adversarial Learning

Pixel-Level Domain Adaptation: A New Perspective for Enhancing Weakly Supervised Semantic Segmentation

A hybrid domain learning framework for unsupervised semantic segmentation.

Dual domain distribution disruption with semantics preservation: Unsupervised domain adaptation for medical image segmentation

Multi-Source Domain Adaptation with Collaborative Learning for Semantic Segmentation

Partial Domain Adaptation on Semantic Segmentation

The Last Mile to Supervised Performance: Semi-Supervised Domain Adaptation for Semantic Segmentation

A Domain Adaptive Semantic Segmentation Method Using Contrastive Learning and Data Augmentation

Diffusion-based Image Translation with Label Guidance for Domain Adaptive Semantic Segmentation

Dual-level Interaction for Domain Adaptive Semantic Segmentation

Double-domain Adaptation Semantics for Retrieval-based Long-term Visual Localization

A Multilevel-Guided Curriculum Domain Adaptation Approach to Semantic Segmentation for High-Resolution Remote Sensing Images.

Dual Super-Resolution Learning for Semantic Segmentation

SPCL: A New Framework for Domain Adaptive Semantic Segmentation via Semantic Prototype-based Contrastive Learning