• Version
• Download 5
• File Size 315.74 KB
• File Count 1
• Create Date 9 December 2024
• Last Updated 9 December 2024

AutoMarineWatcher: Real-Time Maritime Surveillance to Detect Threats using YOLO

Praneeth Vadlapati
VIT-AP University, India
praneeth.18bce7147@vitap.ac.in
ORCID: 0009-0006-2592-2564

Abstract: Maritime security continues to be a significant challenge due to a rise in piracy and helicopter-based hijackings of commercial vessels. While current detection and defense systems focus on large vessels and naval vessels, addressing smaller-scale threats and hijackings remains underexplored. This paper proposes a system called “AutoMarineWatcher,” which is a novel real-time maritime surveillance solution that uses computer vision by leveraging an existing object detection model. The system provides two levels of warnings, allowing crew personnel to observe and initiate subsequent defensive protocols. This allows a reduction in threat response times to enhance security. The system has demonstrated high accuracy by successfully detecting boats, ships, and helicopters. The detections have been successful, and the system proved its effectiveness in both detections and alerts. The performance of the system has been consistent on regular and infrared footage samples. The system ensured the handling of multiple scenarios in regular lighting conditions and low light conditions. The code is available at github.com/Pro-GenAI/AutoMarineWatcher.

Keywords: computer vision, real-time surveillance, maritime surveillance, security, early warning, automated alerts

DOI: 10.55041/ISJEM0023