Channel and Spatial Attention Based Convolutional Neural Network for Monitoring Surface Deformation

Surface deformation is caused through underground mining causes damage to surface buildings and hazards of potential safety and losses of property. Currently, many prediction techniques depended on sampling points, though these techniques avoid to take local and entire spatial features and that affected spatial prediction accuracy outcomes. In this research proposed a Channel and Spatial attention module based Convolutional Neural Network (CS-CNN) for monitoring and predicting surface deformation. The dataset used for monitoring surface deformation is SAR dataset and the prediction is performed by proposed CS-CNN which provides high prediction results. The performance of proposed method is evaluated with performance measure of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Square Error (MSE) and R2. The proposed method attained less RMSE of 0.828, MAE of 1.345, MSE of 1.236 and R2 of 1.036 which is superior than other existing methods like Differential Interferometric Synthetic Aperture Radar (D-InSAR) and Convolution Long Short-Term Memory (ConvLSTM).