September 19, 2025WaterOpen Access

LU-Net: Lightweight U-Shaped Network for Water Body Extraction of Remote Sensing Images

View Full Paper

Ask AI

Bookmark

LU-Net attains the lowest GPU latency of 3.1 MS and second-lowest CPU latency of 36 MS.
On the GID dataset, LU-Net achieves a MIoU of 91.36%, outperforming other methods evaluated.
The model employs depthwise separable convolution to accelerate the decoding process while maintaining accuracy.
Comparative results indicate LU-Net's commendable efficiency across all compared networks in both datasets.

Discussion

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

This method enhances inference speed and accuracy for water body extraction in remote sensing images, suggesting improved real-time applications.

Deng et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

This method enhances inference speed and accuracy for water body extraction in remote sensing images, suggesting improved real-time applications.