As bridge infrastructure continues to age globally, the need for efficient maintenance and digital documentation becomes increasingly urgent. 2D engineering drawings remain a primary source of information for bridge assessment, yet extracting semantic content from these documents remains a manual, time-consuming task. This study proposes an AI-based method for automating the extraction and structuring of information from 2D bridge engineering drawings. The approach integrates YOLOv7-based object detection, Handprint-based OCR, and a post-processing pipeline to identify view regions, recognize textual content, and match view titles with corresponding views. The system outputs semantically annotated views, enabling the generation of machine-readable view directories. Experimental results demonstrate high accuracy in detecting views and recognizing both printed and handwritten text, with minimal manual input. This work contributes to digital workflows in bridge documentation and sets the foundation for integration into broader Bridge Information Modeling (BrIM) systems to support infrastructure lifecycle management.