SwinSeg: Swin transformer and MLP hybrid network for ship segmentation in maritime surveillance system

Accurate identification and segmentation of moving ships to ensure maritime traffic safety has become an important task in maritime intelligent transportation system. With the development of artificial intelligence, maritime surveillance system based on computer vision has been widely studied. However, there are still some problems when applied to the actual maritime scene. For example, the degradation of visible image quality caused by rain, haze, and low illumination leads to a significant reduction in segmentation performance. To improve the performance of ship identification and segmentation under adverse weather conditions, the authors propose SwinSeg, a hybrid network combining Swin Transformer and lightweight multi-layer perceptron (MLP). To address the lack of suitable open-source datasets in the community, the authors have collected and labeled a semantic segmentation dataset of marine ships, named SeaShipsSeg. It consists of 1200 visible marine ship images and covers six common ship types (bulk cargo carrier, container ship, fishing boat, general cargo ship, ore carrier, and passenger ship). In addition, synthetic degraded images are added to the dataset to increase its diversity and improve the generalization ability of the network. The experimental results show that the performance of the authors' method is significantly better than the state-of-the-art (SOTA) methods in terms of segmentation accuracy, robustness, and efficiency under different weather conditions. The dataset is available at https://github.com/GrimreaperZ-creator/SeaShipsSeg.

• Record URL:
  https://doi.org/10.1016/j.oceaneng.2023.114885
• Record URL:
  http://www.sciencedirect.com/science/article/pii/S0029801823012696
• Availability:
  • Find a library where document is available. Order URL: http://worldcat.org/issn/00298018
• Supplemental Notes:
  • © 2023 Elsevier Ltd. All rights reserved. Abstract reprinted with permission of Elsevier.
• Authors:
  • Zhang, Yuqi
  • 0000-0002-9464-5314
  • Li, Chaofeng
  • 0000-0002-3236-3143
  • Shang, Shaopeng
  • Chen, Xinqiang
  • 0000-0001-8959-5108
• Publication Date: 2023-8-1

Language

• English

Media Info

• Media Type: Web
• Features: Figures; References; Tables;
• Pagination: 114885
• Serial:
  • Ocean Engineering
  • Volume: 281
  • Issue Number: 0
  • Publisher: Pergamon
  • ISSN: 0029-8018
  • EISSN: 1873-5258
  • Serial URL: http://www.sciencedirect.com/science/journal/00298018

Subject/Index Terms

• TRT Terms: Computer vision; Intelligent transportation systems; Ships; Surveillance
• Subject Areas: Marine Transportation; Vehicles and Equipment;

Filing Info

• Accession Number: 01885184
• Record Type: Publication
• Files: TRIS
• Created Date: Jun 20 2023 10:09AM