Conference Agenda

Abstract

Monitoring the shipyard production state is of great significance to shipbuilding industry development and coastal resource utilization. Using satellite remote sensing data to monitor the production state of shipyard dynamically has the advantages of high efficiency and objectivity. Further, dock, shipway, assembly area, material storage area and other shipbuilding places are the indispensable part of shipbuilding industry, which can reflect the production activity of the shipyard. This study performs object detection for docks based on high-resolution remote sensing images and deep learning methods. Meanwhile, according to the imaging characteristics of optical remote sensing images and SAR images of shipbuilding places, we used satellite remote sensing data to dynamically monitor the shipyard production state from spatial and time series perspective.

The study firstly uses an object detection network based on the deformable spatial attention module (DSAM), which can be used to detect the docks on high spatial resolution remote sensing image. This network solve the object detection problems caused by the limit of actual docks and the diversity of docks. Secondly, since the backbone of the dock object detection network is with the excellent feature extraction capability for docks, this study connects the backbone with a lightweight status recognition network (Status Head) to determine the dock production status information based on the features extracted from the backbone. Thirdly, we analyzed the backscattered features of shipbuilding places on SAR satellite images, and proposed a method to monitor shipyard production state by using multi-time SAR data. This method can reduce the error caused by insufficient time resolution when using high resolution optical remote sensing data to monitor the production state. Finally, the proposed method can accurately recognize the shipyard production state through experimental verification, which reflects the potential of satellite remote sensing images in shipyard production state monitoring, and also provides a new research thought perspective for other coastal industrial production state monitoring.

.