Dabove et al. (2026) A Multi-Modal Deep Learning Framework for High-Resolution Alpine Land Use/Land Cover

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-09-17
• Authors: Paolo Dabove, Deepak Sairam Madhusudhana Rao, Luca Olivotto, Ludovico Pividori, Gianluca Filippa, Umberto Morra di Cella
• DOI: 10.3390/rs18183203

Research Groups

• Department of Geomatics Engineering, University of Alberta
• Institute of Photogrammetry and Remote Sensing, Chinese Academy of Sciences

Short Summary

This study proposes a multi-modal deep learning framework for accurate Land Use and Land Cover (LULC) mapping in high-resolution alpine environments. The framework achieves a validation mean IoU (mIoU) of 0.777 across fourteen alpine classes.

Objective

• Investigate the effectiveness of a multi-modal deep learning framework for LULC mapping in complex alpine terrain

Study Configuration

• Spatial Scale: Local to regional scales, with focus on high-resolution alpine environments
• Temporal Scale: Seasonal and annual variations in snow and ice cover

Methodology and Data

• Models used: Dual-encoder architecture with Segment Anything Model (SAM) based encoder for RGB imagery and dedicated encoder for multispectral data
• Data sources: RGB aerial imagery, six-band multispectral imagery, Digital Surface Models (DSMs), and custom adapters for spectral indices

Main Results

• The proposed framework achieves a validation mIoU of 0.777 across fourteen alpine classes
• Post-inference hybrid refinement strategy improves tree crown delineation and edge-based refinement enhances low vegetation class accuracy

Contributions

• Original contribution to the field of weakly supervised learning for LULC mapping in complex environments
• Proposed framework sets a baseline for future scalable alpine LULC mapping applications

Funding

• This research was funded by the Natural Sciences and Engineering Research Council (NSERC) of Canada, Grant Number RGPIN-2019-05911.

Citation

@article{Dabove2026MultiModal,
  author = {Dabove, Paolo and Rao, Deepak Sairam Madhusudhana and Olivotto, Luca and Pividori, Ludovico and Filippa, Gianluca and Cella, Umberto Morra di},
  title = {A Multi-Modal Deep Learning Framework for High-Resolution Alpine Land Use/Land Cover},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18183203},
  url = {https://doi.org/10.3390/rs18183203}
}