Automatic CAD Floor Plan Creation with Classification Algorithms

Purpose: Architectural floor plan design is a time-consuming process for architects to develop unique projectsthat meet client requirements. This study proposes an artificialintelligence model to assist and accelerate the design process. Methods: A hybrid deep learning model using Convolu-tional Neural Networks (CNN) and classification algorithms wastrained on a publicly available dataset consisting of 15,000 floorplans, covering 28 categories. The dataset includes elements suchas rooms, doors, windows, stairs, household furniture, and lifts. Results: The model demonstrates high classification accuracyand can generate realistic floor plans. Although the datasetsupports basic plan generation, future extensions may allow formore detailed outputs. Conclusion: This approach provides a novel solution inarchitectural design automation, offering creative inspirationand reducing manual effort. Unlike prior studies, this workdirectly focuses on automatic floor plan generation through deeplearning.