Can anyone give pointers about training PIFu on the Shapenet dataset?

[fariduri]

I want to train the model to recreate 3D car objects by training it with the Shapenetcore dataset. For starters, I know that I have to convert the the meshes to watertight meshes. I tried using ManiFoldPlus but it messes up the UV maps. Is there a way to fix them automatically or do I have to go in and fix it manually one by one using blender? Are there better solutions?

Also how do I configure the training to work optimally and be stable on lower grade hardware? My set up has a Ryzen 5 5600g 16GB of RAM and an RTX 2060 Super.

Thanks!