실 환경에 강인한 2D 이미지 기반의 3D 버텍스 변환 기술 개발

Abstract

This paper develops a robust 2D image-based 3D vertex transformation technology for real-world environments. We introduce a novel approach to estimating 3D models based on human silhouettes extracted from frontal and side images. This research demonstrates the potential applications of real-time 3D modeling, virtual fitting, and health monitoring. Through data augmentation and network training, we have developed a model that is resilient to the diverse postures and appearances of users. This technique particularly offers advantages in effectively responding to complex variations in real-world settings.