This Dress Doesn't Exist

This post was originally published on the Shoprunner Engineering blog here feel free to check it out and at some of the other work our teams are doing. Our ShopRunner Data Science team allows all members to have a quarterly hack week. It is important for data science teams to keep innovating so once per quarter team members are allowed to spend a week working on more speculative projects of their choice. For my 2019 Q3 hack week I decided to build a series of generator models to attempt to create fake products. Generator models are models commonly trained to create realistic images or text based on real world examples. This project may seem fairly outlandish, which it is, but my general idea is that if we can create strong generator models that can capture the diversity of our product catalog then we could use these generators to augment low frequency classes within our catalog for other deep learning projects such as taxonomy classification or attribute tagging.