Training is in progress, results will be published very soon.
We have used Source image of a person along with it's apprel product image as a pair, scraped from Zalando website to train the network. Densepose info obtained using the Detectron2 is used to collect the texture map of the image which is used for the training purpose.