A Comprehensive Survey of Floor Plan Recognition: From Traditional Techniques to Deep Learning Approaches

Floor plan recognition is of great significance in building information modeling (BIM), intelligent applications on buildings and smart cities. With the recent progresses on AI and computer vision, automatic floor plan understanding from 2D architectural plan drawings have become possible. However, prior rules-based approaches have difficulties handling complex irregular shapes while more complicated designs. Adaptive methodologies and semantic cognizance of deep learning techniques do enhance the framework but they need for broad group of labeled facts and additionally, even then they do not affect the chief segment of the communication. Floor plan recognition face challenges in generalization and architect reasoning. We conducted a systematic survey on the research from 2000-2025, review the evolution, advantages and disadvantages, as well as prospects in floor plan recognition methods.