Development of a Marine Mooring Lines Image Dataset for Deep Learning Applications

Abstract

Mooring systems, which use thin lines composed of fiber ropes, steel wires, and chains, are essential to offshore activities. These technologies play a critical role in ensuring that floating units stay in their proper positions throughout operations like offshore gas and oil drilling and production offloading into shuttle storage vessels. Early mooring line failure detection is critical to ensuring the safety and integrity of these activities since it helps avert unanticipated losses such as catastrophic catastrophes and human casualties. Recent developments in deep learning object detection have encouraged prospects for the creation of affordable mooring monitoring solutions. However, a specialized marine images dataset that includes mooring lines is required for the construction and testing of such models, and it is not publicly accessible at this time. Thus, this paper provides insights, design considerations, and general observations for the production of a high-quality marine mooring line images collection in response to the demand for one. Besides, a procedural framework has been designed for creating synthetic images consisting of mooring lines. Additionally, the real-time experimental process in the development of marine mooring lines images dataset for deep learning models that are intended to identify anomalies in mooring lines in offshore deep water has been demonstrated. The purpose of these guidelines is to improve the efficacy of deep learning solutions in determining the anomalies in mooring lines at an early stage by addressing the particular problems and peculiarities of the task.