YOShI: YOLO-ODARIS Ship-detection Interface - An Innovative Approach to On-board Data Analysis for Ship-detection

Abstract

The rapid growth of the demand for low-latency information about time-critical events in the field of remote sensing has become a major focus in the aerospace sector. One solution to this demand, is the possibility to extract key information through automated data analysis on-board the satellite. The availability of state-of-the-art AI-based tools has strongly catalysed the development of such systems. Within this context, the German Aerospace Center (DLR) develops an "On-board Data Analysis And Real-Time Information System" with ship-detection used as a reference application for AI-based on-board object detection. Despite this large sector, there is no available standard approach for image processing and data handling in this field of application. As a result, the "Yolo-ODARIS Ship-detection Interface" is proposed for on-board ship-detection suitable for small- to mid-sized satellites with low on-board processing and optical instrument performance.