| KEYWORD |
Virtual humans and artificial intelligence in the fashion industry
Thesis in external company
keywords AVATAR, COMPUTER ANIMATION, FASHION, MACHINE LEARNING, VIRTUAL AND AUGMENTED REALITY
Reference persons FABRIZIO LAMBERTI
External reference persons Alberto Cannavò
Research Groups GR-09 - GRAphics and INtelligent Systems - GRAINS
Thesis type RESEARCH THESIS
Description The fashion industry is undergoing a digital revolution driven by recent technological improvements. Particularly in design processes and marketing strategies, new solutions have been introduced in recent years. Both market brands and researchers are increasingly investing in technologies like Virtual Fitting Rooms (VFRs) to enrich customer experiences with innovative features, such as Virtual Try-On (VTON) applications, aimed at facilitating shopping. Despite these advancements, challenges persist that could prevent customers from fully leveraging VTON solutions. These challenges include difficulties in recreating high-fidelity avatars of the users and simulating real-time clothing deformations.
In this field, several theses are available. One of the goals is to improve the generation of high-fidelity avatars. Specifically, the thesis requires optimizing an existing framework for the generation of avatars by exploring new solutions aimed at i) better estimating the values of the parametric model describing the avatar, ii) coping with the current limitations of existing frameworks, iii) identifying solutions for generating not only the 3D model but also the textures and fine details of the avatar. Another research direction considers the challenges related to clothes generation and animation. In this respect, the thesis would require the design of new solutions able to i) automatically reconstruct the physical properties of the materials from images and scans, and ii) generate the 3D model of the garment from the sewing patterns. Moreover, the advantages brought by the use of parametric virtual try-on and real-time cloth deformation could be studied.
Suggested readings:
- Cannavo' A.; Pesando R.; Lamberti F. A Framework for Animating Customized Avatars from Monocular Videos in Virtual Try-On Applications. In: International Conference on Extended Reality (2023).
- Gao Qiqi; Takafumi Taketomi. Garment Model Extraction from Clothed Mannequin Scan. In: Computer Graphics Forum. 2023.
- Berthouzoz Floraine, et al. Parsing sewing patterns into 3D garments. In: ACM Transactions on Graphics (TOG). (2013)
- Libao E. I.; Lee M.; Kim S.; Lee S. H. MeshGraphNetRP: Improving Generalization of GNN-based Cloth Simulation. In Proc of the 16th ACM SIGGRAPH Conference on Motion, Interaction and Games. (2023)
See also http://grains.polito.it/work.php
Deadline 15/01/2025
PROPONI LA TUA CANDIDATURA