Maritime surveillance plays a crucial role in ensuring coastal security, safe navigation, environmental protection, and efficient port operations. Conventional monitoring systems such as radar and Automatic Identification Systems (AIS) suffer from several limitations, including high infrastructure costs, dependency on cooperative vessel behavior, and reduced reliability under adverse weather conditions or deliberate transponder shutdowns. These challenges highlight the need for intelligent, cost-effective, and robust alternative surveillance solutions. This paper describes an AI-based system for detecting and monitoring ships that uses deep learning and computer vision to find and classify ships in satellite and aerial images. The suggested system uses the YOLOv8 object detection model [2] to find and classify ships in complicated ocean environments with high accuracy. The processing pipeline consists of preprocessing images, extracting features, finding ships, classifying them, and validating the results after processing. The system’s performance is measured using standard metrics like Precision, Recall, and mean Average Precision (mAP) in different sea states and lighting conditions. The system combines geolocation and weather data to make it more useful for operations. This lets it be aware of its surroundings and automatically send alerts when there are dangerous conditions or unusual vessel behavior, like going too fast or losing an AIS signal in restricted maritime areas. We created an interactive web-based dashboard using Streamlit and Folium that shows detected vessels with real aerial ship photos as circular map markers, vessel counts, geographic locations, Mobile Digital Twin positions, and alert notifications in real time. The proposed system is modular, scalable, and open-source, which means it can easily work with the maritime monitoring systems that are already in place. It is a useful and inexpensive solution for things like monitoring the coast, managing ports, responding to disasters, and keeping an eye on the environment. The AI-based ship recognition framework shows how deep learning, geospatial analytics, and environmental intelligence can be used together to greatly improve maritime situational awareness.

NaviGuardAI functions as a complete maritime ship detection system which uses AI technology to improve situational awareness together with operational efficiency and maritime safety. The system combines deep learning object detection through YOLOv8 with a Mobile Digital Twin tracking system that operates through smartphones and FastAPI backend processing and secure ngrok HTTPS tunnels and interactive geospatial visualization and weather intelligence and automated alert systems to develop a unified and economical maritime decision support system. The framework uses computer vision methods together with mobile sensor combination and geographic analysis and environmental intelligence to achieve maritime monitoring capabilities which exceed traditional detection systems while operating without the need for costly AIS transponder systems. The implementation shows that deep learning models successfully detect and identify vessels through satellite or aerial images with precise accuracy and instant analysis capabilities. The system combines multi-vessel monitoring with interactive geospatial mapping through Folium to transform raw detection data into spatial insights, which help maritime operators to track vessel movements and monitor their operations. The use of real aerial ship photographs as circular map markers enhances visual identification capability, providing operators with photorealistic vessel recognition directly on the interactive dashboard map. The Mobile Digital Twin module demonstrates that a standard smartphone can function as a fully operational vessel transponder without requiring dedicated AIS hardware through its successful implementation. The system achieves real-time vessel tracking across heterogeneous networks including 4G and 5G cellular infrastructure by transmitting live GPS coordinates from the smartphone browser to the FastAPI backend through an ngrok HTTPS tunnel. The Haversine-based distance computation and m/s to knots speed conversion together with the 30-second staleness detection mechanism provide accurate and up-to-date information about mobile position data on the dashboard. This approach directly addresses the dark vessel problem identified in the literature survey, offering a practical and deployable alternative to cooperative AIS-based tracking for resource-constrained maritime environments. The automated alert generation system successfully detects vessel operational violations through its continuous monitoring of vessel operations which it compares to established safety standards. The KPI dashboard bar provides operators with an immediate summary of active alert count, total vessels tracked, moving vessel count, and nearest vessel distance, enabling rapid situational assessment without requiring detailed map inspection. The system gains enhanced operational abilities through its weather intelligence integration which enables safety-critical decision-making by providing environmental data together with vessel tracking information. Unlike The NaviGuardAI framework together with its advanced detection system and real-time GPS tracking and secure network communication and 3D map rendering and behavior detection and environmental monitoring creates a complete system which can be used in actual situations. The system architecture which uses open-source platforms like FastAPI and Streamlit and Folium and YOLOv8 enables system growth and future system updates. The upcoming project will bring together a weather API which will provide real-time weather information based on the Mobile Digital Twin’s current location together with YOLOv8-based visual ship classification which will use camera input and continuous GPS tracking which allows route tracking and assessment and AIS data feed which will support vessel identification and collision danger evaluation which will use vessel movement and heading and distance information. NaviGuardAI proves that it is possible to create an intelligent maritime situational awareness system which requires minimal hardware resources to connect academic research in computer vision with real-world maritime operations. The system establishes a fundamental technical base which will enable upcoming developments in intelligent maritime surveillance that can be used by multiple users throughout various locations.

[1]J. Li, X. Wang, and Z. Zhang, “Ship detection in optical remote sensing images based on deep learning,” IEEE Geoscience and Remote Sensing Letters, vol. 15, no. 7, pp. 1046–1050, Jul. 2018. [2]J. Redmon and A. Farhadi, “YOLOv3: An incremental improvement,” arXiv preprint arXiv:1804.02767, 2018. [3]H. Zhang, Y. Wang, and Q. Li, “Maritime surveillance using computer vision and deep learning techniques,” IEEE Access, vol. 8, pp. 186200–186215, 2020. [4]International Maritime Organization (IMO), “AIS transponders and vessel traffic services,” IMO Publications, London, UK, 2021 [5]S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards real-time object detection with region proposal networks,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 6, pp. 1137–1149, Jun. 2017 Y. Zhang and Q. Wu, “A survey on maritime target detection in remote sensing images,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 146, pp. 1–15, 2018

© 2025 International Journal of Engineering and Techniques (IJET).

Digital Object Identifier (DOI)

IJET assigns a unique DOI to every accepted article via Zenodo for persistent linking and citation. Your article’s DOI: https://doi.org/{{doi}}

Open DOI About Zenodo DOI

Close