Point Cloud Transformers (PCTs.PyTorch)

This repository is forked from qq456cvb/Point-Transformers, and I made the following changes to the code related to the classification task.

1. To adapt Hydar 1.1 & 1.2

• Delete the first line of config/model/*.yaml file: # @package _group_ (this syntax is no longer supported in newer versions of Hydra)

• cls.yaml Modify:

  • defaults added _self_ (required for higher versions)
  • Add hydra.job.chdir = Ture (higher versions already have this property set to False by default)
  • Add custom_output_dir to allow users to customize the output path, so that the same task can be run multiple times.

2. Added test_cls.py

• The purpose of adding this script is to perform deep learning testing and validation experiments separately.
• We conducted test experiments on the ModelNet40 dataset and the program works fine.
• Run with python test_cls.py.

Notes

For other relevant reference details, please check the original repository qq456cvb/Point-Transformers.