I-Corps: Automated Clothing Simulation and Human Avatar Generation Engine

Vanderbilt University, Nashville TN

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is to empower retailers with tools that enable them to make a meaningful reduction in the number of returns generated from e-commerce purchases. It is anticipated that a valuable opportunity will emerge to leverage the newly generated data for better recommendation algorithms, improved forecasting, and enhanced marketing efforts. The retail industry is at a pivotal point because physical locations are becoming increasingly difficult to sustain, leading retailers to develop a stronger online presence. A major challenge preventing fashion retailers from expanding their online presence is the high rate of returns for e-commerce (30-50%) vs. in-store (8.9%) purchases. In addition to the monetary cost of returns ($40 B), there is a significant impact on the environment with billions of pounds of goods going to landfills due to returns. This I-Corps project will deliver accurate human body analogues from minimal user inputs, realistic clothing simulations from non-standardized design and manufacturing files, and an automated integration into credible fit visualizations. The technology provides 1) an automated process for turning design files into photo realistic 3D simulations, 2) an automated process for generating an accurate shadow avatar, and 3) a process by which the avatar and simulation can be combined and multiple meshes layered using position-based dynamics. The technological implementation of this team's approach utilizes deep learning and machine learning algorithms, 3D modeling, mesh generation and mapping, a custom rendering engine, and advanced fabric simulation techniques. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.