Water-HRNet:

• A deep Network for surface water mapping .
• Just a binary classification problem. Water or Not Water.
• Since the training dataset used is too small, the effect of the model is not guaranteed, it is only used to complete my coursework.

-- Model

• We use HRNet as the main model structure（HRNet-W48）.

-- Dataset and Trained Model

• We use the Sentinel-2 image dataset from [WatNet]. Download from [Link].
• Downloading the trained model from [Google Drive] [BaiduNetDisk] code：1111
• The trained model performance:

Acc	mIoU	fwIoU
0.9757	0.9323	0.9528

-- Simple Use

-- Step 1

• clone the rep

git clone https://github.com/faye0078/LULC

• Install dependence

cv2
gdal
matplotlib
numpy
tqdm
Pillow
torch
torchvision

-- Step 2

• Download Sentinel-2 images, and select these six bands. You can also use the other image has same bands(such as landset).

Band	Central Wavelength (µm)	Resolution (m)
Band 2 – Blue	0.490	10
Band 3 – Green	0.560	10
Band 4 – Red	0.665	10
Band 8 – NIR	0.842	10
Band 11 – SWIR	1.610	20
Band 12 – SWIR	2.190	20

• Download the trained model, and move it in dir saved_model.

-- Step 3

• change the input_path and the output_path in test.py
• run the test.py

python test.py

-- Train the model

• download the Sentinel-2 image dataset, and move it in dir dataset.
• change the config.py based on your computer.
• run the train.py

python train.py

-- Acknowledgement

• We thanks the authors for providing some of the code in this repo:
  HRNet and WatNet
• We use the HRNet model code from the former, and use the readTiff, cutImg code from the latter.