In this work, we present a novel scheme of masked autoencoders for point cloud self-supervised learning, termed as Point-MAE. Our Point-MAE is neat and efficient, with minimal modifications based on the properties of the point cloud. In classification tasks, Point-MAE outperforms all the other self-supervised learning methods on ScanObjectNN and ModelNet40. Point-MAE also advances state-of-the-art accuracies by 1.5%-2.3% in the few-shot learning on ModelNet40.
PyTorch >= 1.7.0 < 1.11.0; python >= 3.7; CUDA >= 9.0; GCC >= 4.9; torchvision;
pip install -r requirements.txt
# Chamfer Distance & emd
cd ./extensions/chamfer_dist
python setup.py install --user
cd ./extensions/emd
python setup.py install --user
# PointNet++
pip install "git+https://github.com/erikwijmans/Pointnet2_PyTorch.git#egg=pointnet2_ops&subdirectory=pointnet2_ops_lib"
# GPU kNN
pip install --upgrade https://github.com/unlimblue/KNN_CUDA/releases/download/0.2/KNN_CUDA-0.2-py3-none-any.whl
We use ShapeNet, ScanObjectNN, ModelNet40 and ShapeNetPart in this work. See DATASET.md for details.
| Task | Dataset | Config | Acc. | Download |
|---|---|---|---|---|
| Pre-training | ShapeNet | pretrain.yaml | N.A. | here |
| Classification | ScanObjectNN | finetune_scan_hardest.yaml | 85.18% | here |
| Classification | ScanObjectNN | finetune_scan_objbg.yaml | 90.02% | here |
| Classification | ScanObjectNN | finetune_scan_objonly.yaml | 88.29% | here |
| Classification | ModelNet40(1k) | finetune_modelnet.yaml | 93.80% | here |
| Classification | ModelNet40(8k) | finetune_modelnet_8k.yaml | 94.04% | here |
| Part segmentation | ShapeNetPart | segmentation | 86.1% mIoU | here |
| Task | Dataset | Config | 5w10s Acc. (%) | 5w20s Acc. (%) | 10w10s Acc. (%) | 10w20s Acc. (%) |
|---|---|---|---|---|---|---|
| Few-shot learning | ModelNet40 | fewshot.yaml | 96.3 ± 2.5 | 97.8 ± 1.8 | 92.6 ± 4.1 | 95.0 ± 3.0 |
To pretrain Point-MAE on ShapeNet training set, run the following command. If you want to try different models or masking ratios etc., first create a new config file, and pass its path to --config.
CUDA_VISIBLE_DEVICES=<GPU> python main.py --config cfgs/pretrain.yaml --exp_name <output_file_name>
Fine-tuning on ScanObjectNN, run:
CUDA_VISIBLE_DEVICES=<GPUs> python main.py --config cfgs/finetune_scan_hardest.yaml \
--finetune_model --exp_name <output_file_name> --ckpts <path/to/pre-trained/model>
Fine-tuning on ModelNet40, run:
CUDA_VISIBLE_DEVICES=<GPUs> python main.py --config cfgs/finetune_modelnet.yaml \
--finetune_model --exp_name <output_file_name> --ckpts <path/to/pre-trained/model>
Voting on ModelNet40, run:
CUDA_VISIBLE_DEVICES=<GPUs> python main.py --test --config cfgs/finetune_modelnet.yaml \
--exp_name <output_file_name> --ckpts <path/to/best/fine-tuned/model>
Few-shot learning, run:
CUDA_VISIBLE_DEVICES=<GPUs> python main.py --config cfgs/fewshot.yaml --finetune_model \
--ckpts <path/to/pre-trained/model> --exp_name <output_file_name> --way <5 or 10> --shot <10 or 20> --fold <0-9>
Part segmentation on ShapeNetPart, run:
cd segmentation
python main.py --ckpts <path/to/pre-trained/model> --root path/to/data --learning_rate 0.0002 --epoch 300
Visulization of pre-trained model on ShapeNet validation set, run:
python main_vis.py --test --ckpts <path/to/pre-trained/model> --config cfgs/pretrain.yaml --exp_name <name>
Our codes are built upon Point-BERT, Pointnet2_PyTorch and Pointnet_Pointnet2_pytorch
@inproceedings{pang2022masked,
title={Masked autoencoders for point cloud self-supervised learning},
author={Pang, Yatian and Wang, Wenxiao and Tay, Francis EH and Liu, Wei and Tian, Yonghong and Yuan, Li},
booktitle={Computer Vision--ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23--27, 2022, Proceedings, Part II},
pages={604--621},
year={2022},
organization={Springer}
}
