CSID-GAN: A Customized Style Interior Floor Plan Design Framework Based on Generative Adversarial Network

Abstract

As a revolutionary design approach, generative design could offer promising solutions for intelligent design. Considering the high expenses and poor efficiency inherent in traditional interior design, this paper proposes a customized style interior design (CSID) framework based on Generative Adversarial Network (GAN). The CSID-GAN is a two-stage generative model that could first generate rational interior layout schemes and then personalize the style to achieve desired design outcomes. To this end, various loss functions are incorporated to train the generative model for different design tasks. The dataset-model twining method is iteratively utilized to create more diverse design proposals, enabling the model to extensively learn personalized design concepts. Moreover, for evaluating the design results, a comprehensive assessment system is developed at each stage, and the rationality and applicability of this assessment system were validated. Finally, CSID-GAN is employed in optional style design tests for different house layouts. The experimental results have verified the practical feasibility of CSID-GAN.