Implementasi Pose Estimation untuk Virtual Fitting Room dalam Rekomendasi Ukuran Kaos Secara Realtime

Abstract

The basic human need for clothing serves to protect the body. With the continuous increase in population, the demand for clothing also rises. One of the essential factors to consider when selecting clothing is size, which ensures the wearer feels comfortable and looks attractive. Due to the varying sizes from different manufacturers, the National Standardization Agency (Badan Standarisasi Nasional) has issued a standard for men's t-shirt sizes under SNI 2161:2010. Traditionally, clothing size is determined by trying on garments directly, which can be time-consuming. Therefore, a system has been developed to recommend men's t-shirt sizes based on SNI using image processing and human pose detection with only a smartphone camera, eliminating the need for the previously required Kinect camera, which is relatively expensive. Body measurements are based on the Euclidean Distance method for the width and length of the upper body. Testing was conducted at various distances to evaluate the system's accuracy in providing appropriate clothing size recommendations and to display virtual t-shirt fitting. The Confusion Matrix method was used to calculate the system's performance, achieving the highest accuracy of 92% at a distance of 110 cm.