Research on a ship target detection method in remote sensing images at sea

With the accelerated exploitation of marine resources, there is an increasing demand for marine surveillance, navigation safety and port management, in which ship target detection technology is particularly critical. Traditional ship detection methods rely on manual feature extraction and threshold classification, which are inadequate in the face of environmental changes. In this study, a novel ship detection algorithm is proposed, which integrates YOLOv4, convolutional block attention module and transformer mechanism, not only improves the accuracy and robustness of far-sea ship detection, but also provides a new solution strategy for remote sensing image target detection in complex scenes. The experimental data from the SSDD dataset reveal that the algorithm developed in this research surpasses current leading-edge models in both detection precision and velocity. This is particularly evident in the identification of minor targets and within intricate background scenarios, where the algorithm exhibits marked superiority. © The Authors(s) 2024. Published by Inderscience Publishers Ltd. This is an Open Access Article distributed under the CC BY license.