| Abstract | This study uses deep learning to introduce an automated approach for inspecting a ship’s safety plans. We compared Convolutional Neural Network (CNN) models and selected the most suitable for symbol detection. Due to the challenge of obtaining sufficient training data, we proposed a method for generating virtual data based on existing safety plans, addressing overfitting concerns. A prototype program was developed and tested on sample safety plans, showcasing its effectiveness in reviewing safety plans by accurately identifying various symbols. |
|---|---|
| Publication Date | 2024-06-17 |
In-Chang Yeo, Myung-Il Roh, Min-Chul Kong, Dongki Min, Dongguen Jeong, "An Automated Method for the Review of a Ship’s Safety Plan Based on Deep Learning", Proceedings of ISOPE(International Society of Offshore and Polar Engineers) 2024, Rhodos, Greece, 2024.06.16-21