UNIQ: Unsupervised Point Cloud Quality Assessment Via Natural Statistics Modeling

This repository contains the source code of UNIQ.

For NR mode:

Use: python pristine.py to build the pristine parameters.

Then: python compute_quality.py to evaluate distorted point clouds.

We give the k-fold pristine model in the folder 'perdataset_pristine'. Please read 'README.txt' in the folder to use them.

For RR mode:

The code will be available soon.