SyncFit: AI-powered Mobile eCommerce Application Utilising AR for Virtual Clothing Fitting

With the advent and extensive increase in the use of mobile technology and augmented reality (AR) software, the way people shop for fashion, particularly online, has changed. This paper presents a mobile application, SyncFit, driven by AI to give users the most realistic Virtual Try-On (VTO) experience in real-time. Using ARCore and MediaPipe Pose for detection, SMPL modeling for body shape generation, and GAN for garment rendering. Our system achieves real-time pose estimation, size estimation, and rendering of 3D garments, achieving up to 80 fps on Edge TPU; the size estimation accuracy is 90% based upon deep learning network findings. A small-scale pilot usability study was performed to examine usability, with users showing great satisfaction in aesthetics, fit, and usability. Although effective technical and usability results through statistical analysis were found, further improvements are necessary to stabilize accuracy rates, improve garment hyperrealism, and pose estimation stability as we continue future iterations. SyncFit offers a scalable and immersive solution for personalized fashion retail, bridging the gap between traditional shopping and virtual experiences.