AI enhanced 3D human body shape estimation and garment redressing from clothed 3D scan sequences

Eligibility: UK/EU/International graduates with the required entry requirements

Funding details: Bursary plus tuition fees  (UK/EU/International)

Duration: Full-Time – between three and three and a half years fixed term

Application deadline: 27 May 2024

Interview dates: Will be confirmed to shortlisted candidates

Start date: September 2024

For queries contact: Dr Pengpeng Hu


Circular light trails

Trailblazers

The Trailblazer PhD studentships have been devised and developed by leading early-career researchers at Coventry University.

Introduction

UK fashion retailers, particularly in online retail, have long struggled with the cost of returns. To reduce the number of returns in the fashion industry, 3D body scanning technology is a trend to build digital twins for customers, which makes it possible to fit garments remotely.

Project details

To this end, two main subtasks are necessary: a) an accurate 3D body shape reconstruction, and b) dressing of the obtained body. An accurate 3D body shape can be obtained via 3D body scanning techniques. However, it is inconvenient and an infringement on the right to privacy to scan the undressed body directly. Existing methods of dressing the 3D body can be classified into two main categories: the physically based cloth simulation, and redressing garments from a template to another subject. Redressing scanned garments has shown superior performance in terms of the realistic geometry, compared to the physically based cloth simulation. However, existing methods focus on redressing static scanned garments while it is necessary to redress the 3D scanned garment movement to obtain more realistic and accurate virtual try-on results.

This project focuses on developing AI-based solutions for estimating body shapes and redressing dynamic garments from clothed 3D sequences. This can now be achieved thanks to the rapid developments of deep neural networks and improvements in commodity depth sensors. This project will push the state of the art in specific areas of computer vision related to point cloud acquisition and processing to develop the building blocks of a full “virtual fitting” experience that will be marketed to clients in the fashion industry looking to transition into the digital world, or improve the value generation potential of online marketplaces.

How to apply

To find out more about the project, please contact Dr. Pengpeng Hu

 Queen’s Award for Enterprise Logo
University of the year shortlisted
QS Five Star Rating 2023