September 5, 2025Remote SensingOpen Access

An Enhanced Feature Extraction and Multi-Branch Occlusion Discrimination Network for Road Detection from Satellite Imagery

View Full Paper

Ask AI

Bookmark

The enhanced method achieves an IoU of 64.73 on the DeepGlobe dataset, showing significant improvement in road detection.
Using a multi-directional feature extraction module enhances the network's ability to capture linear road features effectively.
The multi-branch occlusion discrimination module utilizes attention mechanisms to reduce road debris caused by occlusion.
This research highlights the potential for improved road detection methods to facilitate applications in urban planning and emergency response.

Discussion

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

Deep learning demonstrates improved road detection in satellite imagery, suggesting new applications in urban renewal and disaster assessment.

Wu et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

Deep learning demonstrates improved road detection in satellite imagery, suggesting new applications in urban renewal and disaster assessment.