This repository is for Contrastive Embedding Distribution Refinement and Entropy-Aware Attention Network (CEDR) introduced in the following paper:
"Contrastive Embedding Distribution Refinement and Entropy-Aware Attention for 3D Point Cloud Classification"
model.py
and main.py
are available now. Test the pre-trained model:
download ModelNet40, unzip and move modelnet40_ply_hdf5_2048
folder to ./data
put the pre-trained model under ./checkpoints/modelnet
then run (more settings can be modified in main.py
):
python main.py --exp_name=gbnet_modelnet40_eval --model=gbnet --dataset=modelnet40 --eval=True --model_path=checkpoints/modelnet/gbnet_modelnet40.t7
Test the pre-trained model:
training_objectdataset_augmentedrot_scale75.h5
and test_objectdataset_augmentedrot_scale75.h5
files to ./data
./checkpoints/gbnet_scanobjectnn
main.py
):
python main.py --exp_name=gbnet_scanobjectnn_eval --model=gbnet --dataset=ScanObjectNN --eval=True --model_path=checkpoints/gbnet_scanobjectnn/gbnet_scanobjectnn.t7
main.py
Model | Dataset | #Points | Data Augmentation |
Performance on Test Set |
Download Link |
---|---|---|---|---|---|
PointNet++ | ModelNet40 | 1024 | random scaling and translation |
overall accuracy: 93.1% average class accuracy: 91.1% |
google drive |
GBNet | ScanObjectNN | 1024 | random scaling and translation |
overall accuracy: 82.9% average class accuracy: 80.6% |
google drive |
The code is built on GBNet. We thank the authors for sharing the codes. We also thank the Big Data Center of Southeast University for providing the facility support on the numerical calculations in this paper.