Abstract:
This research focuses on developing a machine learning-based recommendation system to choose the optimal floor tile color and design for indoor spaces to improve living spaces' aesthetics, comfort, and functionality. This system is designed for homeowners, architects, and interior designers, enabling them to select tile colors based on three critical architectural factors: the indoor space (large or small), lighting conditions (low or bright), and light temperature (warm or cool). The physical appearance of a space and the mental well-being of its occupants are all influenced by these contextual elements. By exploiting architectural principles that emphasize the interaction of light and space, the system offers personalized tile recommendations that support both visual and emotional design goals that promote relaxation and psychological comfort. Leveraging deep learning, the system classifies space types and lighting conditions to generate personalized tile recommendations that align with visual and emotional design principles. Three deep learning models, EfficientNetB3, ResNet50, and VGG16, were trained using a dataset of 4,540 images, allocated as 70% for training, 15% for validation, and 15% for testing. Among these models, EfficientNetB3 demonstrated the highest classification accuracy for indoor spaces (88.19%), while ResNet50 achieved superior performance in classifying lighting conditions (87.33%) and light temperature (90.67%). The recommendation engine integrates these classifications with expert-approved tile selections, ensuring choices that enhance spatial ambiance while promoting relaxation and psychological comfort. This study bridges the gap between architectural aesthetics and human well-being, offering a smart tool for modern design. The system also holds potential for expansion into home automation and outdoor applications, contributing to the future of intelligent interior design solutions.
