Inference using my own point cloud on S3DIS dataset using PointNet++

[girmzewdie]

Hello, first of all, congratulations on your incredible work. Currently, I have downloaded your repo and successfully trained the semantic segmentation model on the S3DIS dataset following the instructions in the readme, obtaining similar results to S3DIS dataset models. How can I make inferences easily on the trained model using my own point cloud? That is, in what format should you load/save the point cloud? Is it necessary to make a transformation of this before?
Thanks for your time.