This repo contains a re-implementation of the paper Down-to-the-Last-Detail-Virtual-Try-on-with-Detail-Carving with different experiments that i have done while working on the repo

TODO

1. Use Densepose instead of Keypoints
2. Write code to train with single product and model image pair for scenarios when the 2 different images of the same clothes is not available
3. Train on different opensource dataset to check the performance

The code is heavily borrowed from the original repo and pix2pix/pix2pixhd.