Automated 3d Bridge Modeling Using Transformer-Based Semantic Interpretation of Drawing Dimension Annotations

Key Points

• The research aims to develop an automated method for constructing 3D bridge models using drawing dimensions and annotations.
• Incorporated a transformer-based object detection model for analyzing drawing images.
• Utilized optical character recognition to extract dimensions.
• Applied the methodology to inspection drawings for accurate 3D model construction.
• Developed a prototype application based on the proposed method.
• Successfully demonstrated the accurate construction of 3D models from inspection drawings.
• Showed that the proposed methodology enhances efficiency in infrastructure management.
• Created a functional prototype application for bridge inspections.