Object Detection at Sea Using Ensemble Methods Across Spectral Ranges

Abstract Having the option of a temporally unmanned bridge when conditions allow, while maintaining or even enhancing navigational safety, is long term aim in maritime industry. Such system requires excellent perception environment using an array sensors. This paper investigates performance object detection at sea electro-optical sensors relevant spectral ranges and discusses how missed risk minimised for objects within navigation range. Using combination cameras visible, near- far infrared ranges, convolutional neural networks are employed detection. Ensemble techniques suggested to minimise amount detections it shown optimisation confidence thresholds can be used increase performance. The results based on image data from vessels near-coast operation Danish waters.