Virtual fitting solution using 3D human modeling and garments

17th September 2020
Virtual fitting solution using 3D human modeling and garments

The year 2020 has been a challenging year for businesses in a variety of industries. In particular, the pandemic crisis has hit fashion demand hard, and many retailers in high-street are announcing a significant drop in their sales. At this point, a possible idea of virtual fitting rooms has clear benefits to the retailers, such as boosting their sales during pandemic periods and reducing return rates. 

This project consists of developing an automatic system that can generate 3D digital human models from a given 2D human image and fit garments on the rendered 3D human models. This project targets applications such as virtual try-on systems.

Contact: cagriozcinar@gmail.com

References:

  1. DensePose: Dense Human Pose Estimation In The Wild – https://arxiv.org/pdf/1802.00434.pdf
  2. Learning to Transfer Texture from Clothing Images to 3D Humans –  https://arxiv.org/pdf/2003.02050.pdf
  3. Densepose Facebook GitHub repository – https://github.com/facebookresearch/DensePose
  4. Texture transfer model – https://github.com/aymenmir1/pix2surf

Requirement: 

  • Basic understanding of Deep-learning,
  • Strong Python programming skills with knowledge of PyTorch/TensorFlow tools.
  • Ideal candidates should have interest in deep learning in general, and must have the ability to follow new research trends and learn new tools.