Toward More Adaptive Fusion: Sample- and Region-Aware Dynamic Fusion Network for Full- and Missing-Modalities Land Cover Classification
Open Access
- 13 February 2026
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19 (19391404) , 7543-7559
Abstract
In recent years, research on land cover classification (LCC) utilizing optical (OPT) and synthetic aperture radar (SAR) images under full- and missing-modalities scenarios has garnered considerable attention. Nevertheless, existing methods face challenges in sufficiently fusing multimodal features and re constructing missing-modality features. Specifically, conventional fusion methods typically utilize fixed, uniform, and inflexible strategies across all sample pairs, thereby lacking the capacity to select the most targeted fusion method for distinct pairs. Furthermore, these methods rely unconditionally on generated features without considering variations in generation quality across different regions. To address these limitations, this pa per proposes the Sample- and Region-Aware Dynamic Fusion Network (SRDFNet), an architecture that enables more adaptive fusion of features at both the sample and region levels. First, the Multi-Expert Adaptive Fusion (MAF) module is proposed, where each expert functions as an independent fusion unit, allowing dynamic activation of expert combinations tailored to individual sample pairs and thus facilitating adaptive feature fusion at the sample level. Second, the Confidence-based Adaptive Feature Generation (CAFG) module is proposed, which simultaneously produces confidence maps during feature generation to evaluate the reliability of generated features across regions, thereby enabling their refinement. Ultimately, a fusion mechanism in tegrates the refined features with regular modality features, enabling adaptive fusion at the region level. Experiments on the WHU-OPT-SAR and DFC2020 datasets demonstrate that SRDFNet achieves an improvement of 2.30% to 2.90% in mean Intersection over Union (mIoU) under full-modality scenarios, and an enhancement of 0.90% to 3.30% in scenarios with missing-modality. These results substantiate the robustness and effectiveness of the SRDFNet.Keywords
Funding Information
- National Natural Science Foundation of China (42101458)
- National Natural Science Foundation of China (42130112)
- National Key Research and Development Program of China (2024YFC3212200)
This publication has 54 references indexed in Scilit:
- Detecting changes without comparing images: Rules induced change detection in heterogeneous remote sensing imagesISPRS Journal of Photogrammetry and Remote Sensing, 2025
- FlexiSAM: A flexible SAM-based semantic segmentation model for land cover classification using high-resolution multimodal remote sensing imageryISPRS Journal of Photogrammetry and Remote Sensing, 2025
- Land-Unet: A deep learning network for precise segmentation and identification of non-structured land use types in rural areas for green urban space analysisEcological Informatics, 2025
- Multi-task model of adaptive multi-scale feature fusion and adaptive mixture-of-experts for equipment remaining useful life prediction and fault diagnosisExpert Systems with Applications, 2025
- Multidimensional knowledge distillation for multimodal scene classification of remote sensing imagesDigital Signal Processing, 2024
- MGFNet: An MLP-dominated gated fusion network for semantic segmentation of high-resolution multi-modal remote sensing imagesInternational Journal of Applied Earth Observation and Geoinformation, 2024
- Similarity and dissimilarity relationships based graphs for multimodal change detectionISPRS Journal of Photogrammetry and Remote Sensing, 2024
- A dual-stream high resolution network: Deep fusion of GF-2 and GF-3 data for land cover classificationInternational Journal of Applied Earth Observation and Geoinformation, 2022
- MCANet: A joint semantic segmentation framework of optical and SAR images for land use classificationInternational Journal of Applied Earth Observation and Geoinformation, 2022
- Beyond RGB: Very high resolution urban remote sensing with multimodal deep networksISPRS Journal of Photogrammetry and Remote Sensing, 2018