2025A-OS5-8 Crane Operation Status Recognition from Low-Resolution Satellite Imagery for Port Efficiency Assessment

Abstract

This study proposes a framework for analyzing container port operational efficiency by integrating low-resolution satellite imagery with AIS vessel data. A deep learning object detection model was developed to identify quay cranes and classify their operational states as "in-operation" or "standby." Using Sentinel-2 imagery, updated approximately every five days, the model achieved reliable recognition performance despite the limitations of low spatial resolution, particularly improving recall for standby cranes. The identified crane states were combined with AIS vessel dynamics, including berthing time, vessel length, and capacity, enabling berth-level analysis of crane allocation and operational efficiency. Comparative evaluation with previous high-resolution imagery studies highlights the advantages of low-resolution imagery in terms of sample size, temporal frequency, and coverage, offering a cost-effective and scalable solution for long-term monitoring. The results demonstrate the feasibility and practical potential of integrating satellite-derived crane status with AIS data to support efficiency assessment at container ports.