August 18, 2025Scientific ReportsOpen Access

Lightweight multiscale information aggregation network for land cover land use semantic segmentation from remote sensing images

View Full Paper

Ask AI

Bookmark

Achieving 94.8% segmentation accuracy, the model significantly reduces computational overhead for remote sensing tasks.
The encoder-decoder architecture employs dense dilated convolutions to effectively extract multiscale features.
Evaluation using the NITRDrone and UDD6 datasets highlights the model's efficacy compared to traditional approaches.
The compact design enables real-time processing on low-power devices, potentially transforming geospatial analysis.

Discussion

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

This model improves segmentation accuracy in land cover and land use analysis, suggesting benefits for real-time remote sensing applications.

Said et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

This model improves segmentation accuracy in land cover and land use analysis, suggesting benefits for real-time remote sensing applications.