About this document
5 Paper by nusrat.fariya.ete is a document available to read on EtoBox.
The paper presents a lightweight U-Net model for cloud detection in remote sensing images, utilizing depthwise separable convolutions to reduce model size and computational costs. The proposed method achieves over 90% pixel accuracy and is approximately five times faster than traditional U-Net, making it suitable for microsatellites with limited resources. Experimental results demonstrate its effectiveness for both visible and thermal infrared bands in cloud cover assessment.
- Author
- nusrat.fariya.ete
- Language
- EN