Monitoring Urban Growth and Infrastructure Evolution with U-Net Deep Learning Model

Change detection based on remote sensing helps in detecting changes that have been made over time. This paper investigates a deep learning model that can identify changes and qualify the changes that have occurred. We use U-Net model for this purpose. The user uploads images of two different timelines, and the model determines the percentage change that has occurred in the area. We have gathered a data set from multiple regions. The images we collected are VHR images that we analyzed for sufficient learning or model training. We then masked the images and trained the unit model using different classes of masked images. The unit model will assist in forecasting the masked images and, with the help of previous and present time line multimeters, the model will identify changes that have occurred in the region over time.