Code for "Classification of Single-View Object Point Clouds"
This code has been tested with
- python 3.7.12
- pytorch 1.6.0
- CUDA 10.1
- se3cnn [link]
- PartialModelNet40(PM40) dataset [link]
- PartialScanNet15(PS15) dataset [link]
- Pretrained models of PAPNet [link]
- PartialModelNet40(PM40) under different levels of partiality [link]
Command for training PAPNet:
python train.py --dataset pm40 --data_path path_to_pm40_dataset
python train.py --dataset ps15 --data_path path_to_ps15_dataset
Evaluate the results of PAPNet:
python test.py --dataset [pm40|ps15] --data_path path_to_pm40_or_ps15 --model path_to_pretrained_model
python transfer_test_pm2ps.py --pm40_model path_to_pm40_model --ps15_path path_to_ps15_dataset
python transfer_test_ps2pm.py --ps15_model path_to_ps15_model --pm40_path path_to_pm40_dataset
Our code is released under MIT License (see LICENSE file for details).