ChunyangYuan / AGCN

the code of paper "Attention Graph Convolution Network for Image Segmentation in Big SAR Imagery Data"

Attention-Graph-Convolution-Network-for-Image-Segmentation-in-Big-SAR-Imagery-Data

this code implements the method proposed in paper "Attention Graph Convolution Network for Image Segmentation in Big SAR Imagery Data". if it helps you, please kindly cite this paper. https://doi.org/10.3390/rs11212586

How to use?

1. run "python generate_gt.py ", to generate the ground truth data from GT.image
2. run "python main.py" to train and test our model,and print the pixel-level Evaluation Metrics

main.py

train, test our AGCN, and calculate the pixel-level Evaluation Metrics（Kappa, precison, recall, and confusion matrix）. Note: as comparison, this file also provide the code of GCN and GAT. If you want to see the results of GAT or GCN, set two parameters "model_nm" and "model" as GAT or GCN

generate_gt.py

generate the ground truth for training the Network

layers.py

define the layers

model_cnn_shanxipucheng.pth

the trained feature_extraction_net

models9.py

define graph convolution Network

train_feature_extraction_net_Pucheng.py

train the feature_extraction_net

utils9.py

define the function for calculating pixel-level Evaluation Metrics

the code runs in python 2.7

numpy 1.16.2

Pillow 5.4.1

scikit-image 0.14.2

scikit-learn 0.20.3

scipy 1.2.1

segraph 0.5

torch 0.4.0

torchfile 0.1.0

any problem please email me : [e-mail]