Muhammad AsadMuhammad SobanSikander Ali KhanMuhammad Irfan2025-12-172025-12-172025https://escholar.umt.edu.pk/handle/123456789/16718This study introduces a new method that uses Artificial Intelligence (AI) and Machine Learning (ML) models to automatically change old two-dimensional architectural and structural plans into Building Information Modeling (BIM) systems. This method helps solve the usual problems and delays found in manual BIM work, offering clear benefits to the Architecture, Engineering, and Construction (AEC) industry. Significantly, the system employs the Mask R-CNN deep learning model within the Detectron2 framework, trained on a dataset comprising 552 annotated elements from real-world projects, to reliably identify and distinguish primary building components including walls, columns, doors, and beams even when analyzing diverse or low-quality plans. The data augmentation technique was used successfully to make the model resistant to various drawing styles and trained on more than 200 iterations and achieved more than 95 percent accuracy in detection. Seen objects are converted to the IFC standard directly, originating 3D BIM models that may immediately be used in BIM programs. The automatized strategy significantly eliminates the amount of manual work, improves the quality of modelling, and assists sustainable management since it allows performing digital analysis of existing buildings. This study also points out to the scalability of the solution to large-scale projects, collaboration of the team due to standardized models, and usefulness in the energy efficiency planning and renovation planning. In conclusion, this research shows that deep learning can effectively automate the BIM process, providing an efficient and scalable way to convert old building records and support digital transformation in construction. Additionally, this work contributes to Sustainable Development Goals 9 (Industry, Innovation, and Infrastructure) and 11 (Sustainable Cities and Communities), highlighting its importance for sustainable growth.enThesis