An End-to-End Intelligent Review System for Architectural Engineering Drawings

As the demand for high-quality development in the architectural engineering field continues to grow, the quality review of engineering drawings has become particularly critical. Moreover, it continuously impacts the intelligent optimization and decision-making of subsequent processes. This study proposes an end-to-end intelligent review system for architectural engineering drawings based on deep learning. The system integrates various artificial intelligence technologies, including connected component analysis, unsupervised clustering, and fine-tuning. It innovates a multifactor detection cascade algorithm framework and improves the OCR algorithm. Additionally, a logic matching engine is designed for architectural engineering drawing elements, based on clustering and distance. This enables an end-to-end automated drawing review process. The experiments have confirmed that the system can significantly reduce review costs and greatly enhance review efficiency.