GAM : Gradient Attention Module of Optimization for Point Clouds Analysis (AAAI2023)

Reference

@inproceedings{hu2023gam,
  title={GAM : Gradient Attention Module of Optimization for Point Clouds Analysis},
  author={Hu, Haotian and Wang Fanyi and Su Jingwen and Zhou Hongtao and Wang Yaonong and Hu Laifeng and Zhang Yanhao and Zhang Zhiwang}
  booktitle={Association for the Advance of Artificial Intelligence, AAAI},
  year={2023}
}

Environment

torch==1.9.0
cuda==11.1.0
cudnn==8.2.1

Classification (ModelNet40)

Data Preparation

Download alignment ModelNet here and save in data/modelnet40_normal_resampled/.

Run

You can run different modes with following codes.

• If you want to use offline processing of data, you can use --process_data in the first run. You can download pre-processd data here and save it in data/modelnet40_normal_resampled/.

# ModelNet40
## Select different models in ./models 

## e.g., pointnet2_ssg without normal features
python train_classification.py --model GAM_cls_ssg --log_dir GAM_cls_ssg
python test_classification.py --log_dir GAM_cls_ssg

## e.g., pointnet2_ssg with normal features
python train_classification.py --model GAM_cls_ssg --use_normals --log_dir GAM_cls_ssg_normal
python test_classification.py --use_normals --log_dir GAM_cls_ssg_normal

## e.g., pointnet2_ssg with uniform sampling
python train_classification.py --model GAM_cls_ssg --use_uniform_sample --log_dir GAM_cls_ssg_fps
python test_classification.py --use_uniform_sample --log_dir GAM_cls_ssg_fps

Performance

Part Segmentation (ShapeNet)

Data Preparation

Download alignment ShapeNet here and save in data/shapenetcore_partanno_segmentation_benchmark_v0_normal/.

Run

## Check model in ./models 
## e.g., pointnet2_msg
python train_partseg.py --model GAM_part_seg_msg --normal --log_dir GAM_part_seg_msg
python test_partseg.py --normal --log_dir GAM_part_seg_msg

Performance

Semantic Segmentation (S3DIS)

Data Preparation

Download 3D indoor parsing dataset (S3DIS) here and save in data/s3dis/Stanford3dDataset_v1.2_Aligned_Version/.

cd data_utils
python collect_indoor3d_data.py

Processed data will save in data/s3dis/stanford_indoor3d/.

Run

## Check model in ./models 
## e.g., pointnet2_ssg
python train_semseg.py --model pointnet2_sem_seg --test_area 5 --log_dir pointnet2_sem_seg
python test_semseg.py --log_dir pointnet2_sem_seg --test_area 5 --visual

Performance