Interior design display based on NST algorithm and AR

Abstract

This study proposes an interior design display model based on a neural style transfer algorithm and augmented reality technology to address the current time-consuming, single-function, and high information requirements of interior design display methods. This model combines virtual reality overlay technology to enhance users’ visual and auditory interaction experience, reasonably allocates the weights of the style loss function and content loss function, and reduces spatial loss. In the experiment, the model training achieved a loss value of 162,091.33 and a space loss rate of more than 1%. In practical applications, a reality reproduction score of 8.7 was achieved, with a peak signal-to-noise ratio higher than 50 dB, indicating a relatively small loss of image quality. Furthermore, the click-through rate of users has increased by 0.115, and the purchase conversion rate has increased by 0.270, indicating that this model has significant effectiveness and advancement in interior design displays.