An explainable AI-based approach for identifying interior design style principles

Abstract

Interior design is crucial in shaping the characteristics of built environments. While designers often draw on expertise and intuition to define interior-styles such as Natural or Modern, formalized principles to systematically analyze these styles remain limited. This study presents a data-driven approach for identifying design principles that differentiate interior-styles by recognizing furnishing types and analyzing their color and material (FCM) using explainable AI (XAI). First, furnishing objects with FCM information were identified from user-generated interior images from two culturally distinct datasets—South Korea (N = 2,979) and the U.S. (N = 2,000)—using object, material, and color recognition. Second, classification models were built to distinguish interior-styles, forming a basis for analyzing how FCM combinations contribute to style differentiation. Third, a Style Explanation Value (SEV) was proposed to interpret the impact of FCM combinations on classification. The method identified distinctive and diverse FCM combinations, interpreted theoretically to reveal cultural variation. Additionally, simplified computation was introduced for practical use. To the best of our knowledge, this is the first study to use XAI in interior design analysis, offering object-level interpretation of how FCM combinations define stylistic identity.