On this page you can find code that will allow you to explore different 2D-to-2D and 2D-to-3D image editing techniques, as well as collect simple datasets. Note that Paperspace is a cloud computing platform that allows you to access GPUs, which are necessary to run deep learning models efficiently. A tutorial on how to use the code in this repository with Paperspace is given in the following section.
For step by step instructions on using Paperspace, please see the following:
Uploading your dataset to and downloading your model outputs from Paperspace
Running and training models using the Gradient Experiment Builder
To access the tutorials for how to run/train a variety of different models using the Paperspace Gradient Experiment Builder:
For a step-by-step walk through of the above folders and files and the code they contain: