October 15, 2025MathematicsOpen Access

A Multi-Modal Attention Fusion Framework for Road Connectivity Enhancement in Remote Sensing Imagery

View Full Paper

Ask AI

Bookmark

The proposed framework significantly improves road connectivity in remote sensing imagery, overcoming occlusion challenges caused by environmental factors.
Experimental results show improvements in precision, recall, F1-score, and mIoU on benchmark datasets like DeepGlobe and Massachusetts.
The dual-stream encoder architecture processes RGB images and road masks, capturing spatial and semantic information effectively.
Overall, this approach enhances road completeness and filtration of noise, demonstrating its potential for real-world applications.

Discussion

No takes yet. Share an insight, caveat, or question.

Proposed framework improves road extraction accuracy in remote sensing imagery, highlighting deep learning advancements.

Yuan et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark