A Modified D-LinkNet for Water Extraction from High-Resolution Remote Sensing

Comparative Performance Analysis of Simple U-Net, Residual Attention U-Net, and Vgg16-U-Net for Inventory Inland Water Bodies

“…The proposed method obtained a 0.842 metric value in terms of the IoU metric. Chang et al (2022) proposed modified U-Net with residual mechanism and attention mechanism in encoder section based on PMS1 remote sensing data of GF2 satellite. The authors achieved good result (i.e., IoU = 0.9270).…”

“…The code used in this study is publicly available on our Gitlab repository (https://git.gfz-potsdam.de/ali/remotesensing-hida). (Zhong et al, 2022) 0.862 --Modified Encoder-Decoder (Zhang and Wang, 2019) 0.984 --DensePPM (Xiang et al, 2023) 0.842 --Res2U-Net (Chang et al, 2022) 0.9270 --ResNet50 (An and Rui, 2022) 0.9781 --U-Net (Ch et al, 2022) 0.89 --…”

A Learning Framework With Multispectral Band-Differentiated Encoding for Remote Sensing Water Body Detection

“…In [20], a framework based on Attention U-Net and LinkNet ingeniously combines prior knowledge generated by numerical simulators to predict the maximum water levels of floods and the terrain deformations caused by floods and debris flows. A polarization self-attention mechanism (PSA) was incorporated into D-LinkNet by [21] to reduce information loss during dimensionality reduction, and its outstanding performance was validated on a dataset constructed from Gaofen-2 satellite remote sensing images. In addition to the classic neural networks, which have been repeatedly proven to have excellent scalability and enormous potential, Hong et al [22] have integrated cutting-edge generative pre-training Transformer (GPT) technology with remote sensing detection.…”