VESSELimg: A Large UAV-Based Vessel Image Dataset for Port Surveillance

In the domain of maritime security and safety, the surveillance and monitoring of vessels in ports are crucial for safeguarding against potential threats and ensuring the overall integrity of port operations. Current port surveillance methods, including radar systems, CCTV networks, and AIS, exhibit limitations, necessitating innovative solutions. The integration of drone-captured imagery holds promise, particularly for detecting non-cooperative vessels that may evade traditional surveillance methods. This paper presents VESSELimg dataset, a meticulously gathered and annotated collection of drone-captured images within port environments. This dataset serves as a benchmark for training and validating deep learning models, specifically tailored for automatic vessel detection, covering diverse vessel types. The implementation of a YOLO-based deep learning model for real-time inference demonstrates the practical applicability of the dataset, underscoring its potential for enhancing security and safety measures in port environments.