Two-Stage Conversion of GA Drawings into 3D Model Using Deep Learning
Year: 2025
Type: conference-paper
Abstract: This study explores the conversion of 2D General Arrangement (GA) drawings of ships into 3D arrangement models using deep learning. We start investigating the current status of key AI/deep learning tools available, such as ChatGPT, YOLO, DETR, and SAM, for our case. Our main study is performed using SAM, by a two-stage process: firstly, performing segmentation and extracting data from GA images. Secondly, combining data into a 3D model of a ship. Compared to the other tools or methods, which failed to function properly and had limitations, our approach successfully operated in recognizing GAs and extracting data. The approach was validated using GAs of commercial vessels, especially tankers. These results are expected to improve the utilization of 2D drawings currently used in shipyards and increase the connectivity and integration between early-stage ship design and later stages of work.
Authors: Ha, Jisang, Gaspar, Henrique
Institution: Norwegian University of Science and Technology
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Funder: European Commission
Award: SMART EUROPEAN SHIPBUILDING
OpenAlex ID: W7083671552
FWCI:
Cites:
Cited by:
Related to: