Generating 3D Model for Human Body Shapes from 2D images using Deep Learning

摘要

Online shopping started to grow all over the world in the last decade, predominantly in the last year due to COVID19. As a result of the lockdown and because many people did not want to take the risk of visiting stores not to be infected, customers direct their full attention to online shopping. That affected both customers and stores’ owners since customers always spend a long time trying to pick the right size and fail most of the time causing high percentage of returns which affects stores’ sales. In this paper part of the solution is introduced by collecting 5 anthropometric parameters from the user and the rest is predicted using an imputation strategy. Feature selection techniques are also applied and a 3D model for the human body is presented to the user on a mobile application.