Vessel Bearing Estimation Using Visible and Thermal Imaging

Ajinkya Gorad*, Syeda Sakira Hassan, Simo Särkkä

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

66 Downloads (Pure)


Maritime awareness and autonomous navigation can be enabled by state-of-the-art deep learning methods, by monitoring and tracking the position of maritime vessels. In our experiment, we acquire ship dataset from Megastar cruise ferry campaigns in Baltic Sea. We detect the nearby vessels using visible and infrared imaging sensors and object detectors You-Only-Look-Once v5 (YOLOv5), and Detectron2 RCNN network and use that information along with DeepSORT method to track the position of the vessel. We obtain the bearing of the vessels detected from both infrared and visible image sequences of ship and fuse them using a Kalman filter (KF) and Rauch-Tung-Striebel (RTS) smoother. We then compare the result to the bearing obtained from the automatic identification system (AIS) of the vessel, and also compare it among object detectors. We obtained a root-mean-square error of 0.17 in vessel bearing tracking as compared to AIS.

Original languageEnglish
Title of host publicationImage Analysis - 23rd Scandinavian Conference, SCIA 2023, Proceedings
EditorsRikke Gade, Michael Felsberg, Joni-Kristian Kämäräinen
Number of pages9
ISBN (Electronic)978-3-031-31438-4
ISBN (Print)978-3-031-31437-7
Publication statusPublished - 2023
MoE publication typeA4 Conference publication
EventScandinavian Conference on Image Analysis - Levi, Kittilä, Finland
Duration: 18 Apr 202321 Apr 2023
Conference number: 23

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume13886 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


ConferenceScandinavian Conference on Image Analysis
Abbreviated titleSCIA


  • Bearing estimation
  • Maritime awareness
  • Ship detection


Dive into the research topics of 'Vessel Bearing Estimation Using Visible and Thermal Imaging'. Together they form a unique fingerprint.

Cite this