Ship Detection From Optical Satellite Images With Deep Learning

Creative Commons License

Kartal M., Duman O.

9th International Conference on Recent Advances in Space Technologies (RAST), İstanbul, Turkey, 11 - 14 June 2019, pp.479-484 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/rast.2019.8767844
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.479-484
  • Keywords: ship detection, optical satellite image, deep learning, tensorflow, python, artificial neural networks
  • Istanbul Technical University Affiliated: Yes


The effective use and control of maritime routes in the commercia/military area is an increasing and important need for states. There are some motivations for this purpose: Safe traffic of the ships in narrow canals, avoiding illegal usage of anchoring areas of ships, monitoring fishing activities to avoid illegal fishing or protect fish population, detection of lost ships, boats or debris in the ocean, detection and identification of warships (intelligence, defense, offensive, etc.). This paper proposed an open source, fast running ship detection system from optical satellite images with the deep learning algorithm. The system does not need any comprehensive hardware, even can work on an average laptop. Tensorflow Object Detection Application Programming Interface (API) is trained by optical satellite images with ships and used as object detection API.