Coastline Change Monitoring Based on Sentinel-2 Image Deep Learning

Based on Sentinel-2 images, this paper carried out the monitoring of coastline changes in Lushunkou area. In view of the problems that the previous artificial field survey coastline cannot reach the harsh environment, there is no coastline classification data set that can be used for deep learning, and the artificial interpretation part of coastline monitoring accounts for a large proportion and has low timeliness, this paper proposes an intelligent interpretation method of coastline changes taking Lushunkou District as an example, constructs a coastline classification data set that can be used for deep learning, and builds a PSPNet semantic segmentation platform for automatic coastline interpretation. The artificial and natural coastlines of Lushunkou District from 2016 to 2021 were extracted, and the coastline changes of Lushunkou District in 6 years were analyzed. The results show that the coastline of Lushunkou District has not changed significantly in space, and the changes of natural and artificial coastline are in line with the background of 2016-2021. The technical process and implementation method studied in this paper can provide reference for the marine department of Lushunkou District to evaluate the effectiveness of its natural coastline restoration work, and provide experience for other coastal areas to monitor coastline changes.