Repo for "Benchmarking Robustness of 3D Point Cloud Recognition against Common Corruptions" https://arxiv.org/abs/2201.12296

Jiachen Sun

Last update: Dec 29, 2022

Related tags

Deep Learning benchmark computer-vision deep-learning pytorch regularization data-augmentation robustness point-cloud-processing ml-safety corruption-robustness

Overview

Benchmarking Robustness of 3D Point Cloud Recognition against Common Corruptions

This repo contains the dataset and code for the paper Benchmarking Robustness of 3D Point Cloud Recognition against Common Corruptions by Jiachen Sun et al. This codebase is based on SimpleView, and we thank the authors for their great contributions.

ModelNet40-C

More visualizations can be found here.

Download ModelNet40-C from Google Drive.

Download ModelNet40-C using our provided script.

Download ModelNet40-C from Zenodo.

ModelNet40-C Leaderboard

Architecture+Data Augmentation Leaderboard

Architecture	Data Augmentation	Corruption Error Rate (%)	Clean Error Rate (%)	Checkpoint
PCT	PointCutMix-R	16.3	7.2	checkpoint
PCT	PointCutMix-K	16.5	6.9	checkpoint
DGCNN	PointCutMix-R	17.3	6.8	checkpoint
PCT	RSMix	17.3	6.9	checkpoint
DGCNN	PointCutMix-K	17.3	7.4	checkpoint
RSCNN	PointCutMix-R	17.9	7.6	checkpoint
DGCNN	RSMix	18.1	7.1	checkpoint
PCT	PGD Adv Train	18.4	8.9	checkpoint
PointNet++	PointCutMix-R	19.1	7.1	checkpoint
PointNet++	PointMixup	19.3	7.1	checkpoint
PCT	PointMixup	19.5	7.4	checkpoint
SimpleView	PointCutMix-R	19.7	7.9	checkpoint
RSCNN	PointMixup	19.8	7.2	checkpoint
PointNet++	PointCutMix-K	20.2	6.7	checkpoint

We allow users to directly download all pre-trained models with every data augmentation method here.

Architecture Leaderboard

Architecture	Corruption Error Rate (%)	Clean Error Rate (%)	Checkpoint
CurveNet	22.7	6.6	checkpoint
PointNet++	23.6	7.0	checkpoint
PCT	25.5	7.1	checkpoint
GDANet	25.6	7.5	checkpoint
DGCNN	25.9	7.4	checkpoint
RSCNN	26.2	7.7	checkpoint
SimpleView	27.2	6.1	checkpoint
PointNet	28.3	9.3	checkpoint
PointMLP	31.9	6.3	checkpoint
PointMLP-Elite	32.4	7.2	checkpoint

More models' results coming soon ......

We allow users to directly download all pre-trained models with standard training here.

Getting Started

First clone the repository. We would refer to the directory containing the code as ModelNet40-C.

git clone --recurse-submodules [email protected]:jiachens/ModelNet40-C.git

Requirements

The code is tested on Linux OS with Python version 3.7.5, CUDA version 10.0, CuDNN version 7.6 and GCC version 5.4. We recommend using these versions especially for installing pointnet++ custom CUDA modules.

[02-23-2022] The updated codes are tested on Python version 3.7.5, CUDA version 11.4, CuDNN version 8.2 and GCC version 7.5 with the latest torch and torchvision libs, but we still suggest the original setup in case of any unstable bugs.

Install Libraries

We recommend you first install Anaconda and create a virtual environment.

conda create --name modelnetc python=3.7.5

Activate the virtual environment and install the libraries. Make sure you are in ModelNet40-C.

conda activate modelnetc
pip install -r requirements.txt
conda install sed  # for downloading data and pretrained models

For PointNet++, we need to install custom CUDA modules. Make sure you have access to a GPU during this step. You might need to set the appropriate TORCH_CUDA_ARCH_LIST environment variable depending on your GPU model. The following command should work for most cases export TORCH_CUDA_ARCH_LIST="6.0;6.1;6.2;7.0;7.5". However, if the install fails, check if TORCH_CUDA_ARCH_LIST is correctly set. More details could be found here.

Third-party modules pointnet2_pyt, PCT_Pytorch, emd, and PyGeM can be installed by the following script.

./setup.sh

Download Datasets Including ModelNet40-C and Pre-trained Models

Make sure you are in ModelNet40-C. download.sh script can be used for downloading all the data and the pretrained models. It also places them at the correct locations.

To download ModelNet40 execute the following command. This will download the ModelNet40 point cloud dataset released with pointnet++ as well as the validation splits used in our work.

./download.sh modelnet40

To generate the ModelNet40-C dataset, please run:

python data/process.py
python data/generate_c.py

NOTE that the generation needs a monitor connected since Open3D library does not support background rendering.

We also allow users to download ModelNet40-C directly. Please fill this Google form while downloading our dataset.

./download.sh modelnet40_c

To download the pretrained models with standard training recipe, execute the following command.

./download.sh cor_exp

To download the pretrained models using different data augmentation strategies, execute the following command.

./download.sh runs

New Features

[02-23-2022]

We include PointMLP-Elite and GDANet in our benchmark

[02-18-2022]

We include CurveNet and PointMLP in our benchmark

[01-28-2022]

We include Point Cloud Transformer (PCT) in our benchmark
ModelNet40-C/configs contains config files to enable different data augmentations and test-time adaptation methods
ModelNet40-C/aug_utils.py contains the data augmentation codes in our paper
ModelNet40-C/third_party contains the test-time adaptation used in our paper

Code Organization In Originial SimpleView

ModelNet40-C/models: Code for various models in PyTorch.
ModelNet40-C/configs: Configuration files for various models.
ModelNet40-C/main.py: Training and testing any models.
ModelNet40-C/configs.py: Hyperparameters for different models and dataloader.
ModelNet40-C/dataloader.py: Code for different variants of the dataloader.
ModelNet40-C/*_utils.py: Code for various utility functions.

Running Experiments

Training and Config files

To train or test any model, we use the main.py script. The format for running this script is as follows.

python main.py --exp-config <path to the config>

The config files are named as <protocol>_<model_name><_extra>_run_<seed>.yaml (<protocol> ∈ [dgcnn, pointnet2, rscnn]; <model_name> ∈ [dgcnn, pointnet2, rscnn, pointnet, simpleview] ). For example, the config file to run an experiment for PointNet++ in DGCNN protocol with seed 1 dgcnn_pointnet2_run_1.yaml. To run a new experiment with a different seed, you need to change the SEED parameter in the config file. All of our experiments are done based on seed 1.

We additionally leverage PointCutMix: configs/cutmix, PointMixup: configs/mixup, RSMix: configs/rsmix, and PGD-based adversarial training configs/pgd as the training-time config files.

For example, to train PCT with PointCutMix-R, please use the following command:

python main.py --exp-config configs/cutmix/pct_r.yaml

Evaluate a pretrained model

We provide pretrained models. They can be downloaded using the ./download.sh cor_exp and ./download.sh runs commands and are stored in the ModelNet40-C/runs (for data augmentation recipes) and ModelNet40-C/cor_exp (for standard trained models) folders. To test a pretrained model, the command is of the following format.

Additionally, we provide test-time config files in configs/bn and configs/tent for BN and TENT in our paper with the following commands:

python main.py --entry test --model-path <cor_exp/runs>/<cfg_name>/<model_name>.pth --exp-config configs/<cfg_name>.yaml

We list all the evaluation commands in the eval_cor.sh, eval_og.sh, eval_tent_cutmix.sh scripts. Note that in eval_cor.sh it is expected that pgd with PointNet++, RSCNN, and SimpleView do not have outputs since they do not fit the adversarial training framework. We have mentioned this in our paper.

Citation

Please cite our paper and SimpleView if you use our benchmark and analysis results. Thank you!

@article{sun2022benchmarking,
      title={Benchmarking Robustness of 3D Point Cloud Recognition Against Common Corruptions}, 
      author={Jiachen Sun and Qingzhao Zhang and Bhavya Kailkhura and Zhiding Yu and Chaowei Xiao and Z. Morley Mao},
      journal={arXiv preprint arXiv:2201.12296},
      year={2022}
}

@article{goyal2021revisiting,
  title={Revisiting Point Cloud Shape Classification with a Simple and Effective Baseline},
  author={Goyal, Ankit and Law, Hei and Liu, Bowei and Newell, Alejandro and Deng, Jia},
  journal={International Conference on Machine Learning},
  year={2021}
}

References

[1] Zhang, Jinlai, et al. "PointCutMix: Regularization Strategy for Point Cloud Classification." arXiv preprint arXiv:2101.01461 (2021).

[2] Chen, Yunlu, et al. "Pointmixup: Augmentation for point clouds." Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part III 16. Springer International Publishing, 2020.

[3] Lee, Dogyoon, et al. "Regularization Strategy for Point Cloud via Rigidly Mixed Sample." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021.

[4] Sun, Jiachen, et al. "Adversarially Robust 3D Point Cloud Recognition Using Self-Supervisions." Advances in Neural Information Processing Systems 34 (2021).

[5] Schneider, Steffen, et al. "Improving robustness against common corruptions by covariate shift adaptation." arXiv preprint arXiv:2006.16971 (2020).

[6] Wang, Dequan, et al. "Tent: Fully test-time adaptation by entropy minimization." arXiv preprint arXiv:2006.10726 (2020).

[7] Qi, Charles R., et al. "Pointnet: Deep learning on point sets for 3d classification and segmentation." Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.

[8] Qi, Charles R., et al. "Pointnet++: Deep hierarchical feature learning on point sets in a metric space." arXiv preprint arXiv:1706.02413 (2017).

[9] Liu, Yongcheng, et al. "Relation-shape convolutional neural network for point cloud analysis." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2019.

[10] Wang, Yue, et al. "Dynamic graph cnn for learning on point clouds." Acm Transactions On Graphics (tog) 38.5 (2019): 1-12.

[11] Goyal, Ankit, et al. "Revisiting Point Cloud Shape Classification with a Simple and Effective Baseline." arXiv preprint arXiv:2106.05304 (2021).

[12] Guo, Meng-Hao, et al. "Pct: Point cloud transformer." Computational Visual Media 7.2 (2021): 187-199.

[13] Xiang, Tiange, et al. "Walk in the cloud: Learning curves for point clouds shape analysis." Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021.

[14] Ma, Xu, et al. "Rethinking Network Design and Local Geometry in Point Cloud: A Simple Residual MLP Framework." arXiv preprint arXiv:2202.07123 (2022).

Comments

PGD-based adversarial training for Pointnet++

Hi, jiachen. In the paper, it says " PointNet++ and RSCNN leverage ball queries to find neighboring points, which will hinder the gradients from backward propagating to the original point cloud, making adversarial training inapplicable." I am wondering why the ball query operation hinder the gradients from backward propagating. If so, is the optimization of pointnet++ in standard training mode also be influenced? This makes me confused, hope to receive your reply!

opened by xingbw 5
Using the code on Kitti datasets.

Hello, Thanks for your great work！ I would like to ask why the layered phenomenon as shown in the figure occurs when the code is used to process Kitti datasets. Could you guide me here? Appreciate your work and the help. Thanks a lot！

opened by kkkcx 3
About implementation for Metrics

Hi authors, thanks for your solid work!

In the given codebase I have not found the implementation of evaluation metrics for modelnet40-c, i.e. error rate (ER) or class-wise mean error rate (mER). So where can I find the evaluation metric code in the repo?

And in the Metric section of the paper, is the average operation missing here?

opened by xingbw 2
batchnorm error

when I run python main.py --entry test --model-path runs/cutmix_r_pct_run_1/model_best_test.pth --exp-config configs/tent_cutmix/pct.yaml

I got this error:

ADAPT: ITER: 10 METHOD: tent AUG: BETA: 1.0 MIXUPRATE: 0.4 NAME: none PROB: 0.5 DATALOADER: MODELNET40_C: corruption: uniform severity: 1 test_data_path: ./data/modelnet40_c/ MODELNET40_DGCNN: num_points: 1024 test_data_path: ./data/modelnet40_ply_hdf5_2048/test_files.txt train_data_path: ./data/modelnet40_ply_hdf5_2048/train_files.txt valid_data_path: ./data/modelnet40_ply_hdf5_2048/train_files.txt MODELNET40_PN2: num_points: 1024 test_data_path: ./data/modelnet40_ply_hdf5_2048/test_files.txt train_data_path: ./data/modelnet40_ply_hdf5_2048/train_files.txt valid_data_path: ./data/modelnet40_ply_hdf5_2048/train_files.txt MODELNET40_RSCNN: data_path: ./data/ num_points: 1024 test_data_path: test_files.txt train_data_path: train_files.txt valid_data_path: train_files.txt batch_size: 32 num_workers: 0 EXP: DATASET: modelnet40_c EXP_ID: c_pct_run_1 LOSS_NAME: smooth METRIC: acc MODEL_NAME: pct OPTIMIZER: pct SEED: 1 TASK: cls EXP_EXTRA: no_test: False no_val: True save_ckp: 25 test_eval_freq: 1 val_eval_freq: 1 MODEL: MV: backbone: resnet18 feat_size: 16 PN2: version_cls: 1.0 RSCNN: ssn_or_msn: True TRAIN: early_stop: 300 l2: 0.0001 learning_rate: 0.0001 lr_clip: 1e-05 lr_decay_factor: 0.5 lr_reduce_patience: 10 num_epochs: 300 Pct( (model): Pct( (conv1): Conv1d(3, 64, kernel_size=(1,), stride=(1,), bias=False) (conv2): Conv1d(64, 64, kernel_size=(1,), stride=(1,), bias=False) (bn1): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (bn2): BatchNorm1d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (gather_local_0): Local_op( (conv1): Conv1d(128, 128, kernel_size=(1,), stride=(1,), bias=False) (conv2): Conv1d(128, 128, kernel_size=(1,), stride=(1,), bias=False) (bn1): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (bn2): BatchNorm1d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) ) (gather_local_1): Local_op( (conv1): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False) (conv2): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False) (bn1): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (bn2): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) ) (pt_last): Point_Transformer_Last( (conv1): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False) (conv2): Conv1d(256, 256, kernel_size=(1,), stride=(1,), bias=False) (bn1): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (bn2): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (sa1): SA_Layer( (q_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (k_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (v_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (trans_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (after_norm): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (act): ReLU() (softmax): Softmax(dim=-1) ) (sa2): SA_Layer( (q_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (k_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (v_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (trans_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (after_norm): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (act): ReLU() (softmax): Softmax(dim=-1) ) (sa3): SA_Layer( (q_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (k_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (v_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (trans_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (after_norm): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (act): ReLU() (softmax): Softmax(dim=-1) ) (sa4): SA_Layer( (q_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (k_conv): Conv1d(256, 64, kernel_size=(1,), stride=(1,), bias=False) (v_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (trans_conv): Conv1d(256, 256, kernel_size=(1,), stride=(1,)) (after_norm): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (act): ReLU() (softmax): Softmax(dim=-1) ) ) (conv_fuse): Sequential( (0): Conv1d(1280, 1024, kernel_size=(1,), stride=(1,), bias=False) (1): BatchNorm1d(1024, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (2): LeakyReLU(negative_slope=0.2) ) (linear1): Linear(in_features=1024, out_features=512, bias=False) (bn6): BatchNorm1d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (dp1): Dropout(p=0.5, inplace=False) (linear2): Linear(in_features=512, out_features=256, bias=True) (bn7): BatchNorm1d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True) (dp2): Dropout(p=0.5, inplace=False) (linear3): Linear(in_features=256, out_features=40, bias=True) ) ) Recovering model and checkpoint from ~/ModelNet40-C/runs/cutmix_r_pct_run_1/model_best_test.pth Adaptation Done ... N/A% (0 of 78) | | Elapsed Time: 0:00:00 ETA: --:--:--Adaptation Done ... 1% (1 of 78) | | Elapsed Time: 0:00:01 ETA: 0:01:52Adaptation Done ... 2% (2 of 78) | | Elapsed Time: 0:00:03 ETA: 0:01:55Adaptation Done ... 3% (3 of 78) | | Elapsed Time: 0:00:04 ETA: 0:01:56Adaptation Done ... 5% (4 of 78) |# | Elapsed Time: 0:00:06 ETA: 0:01:52Adaptation Done ... 6% (5 of 78) |# | Elapsed Time: 0:00:07 ETA: 0:01:51Adaptation Done ... 7% (6 of 78) |# | Elapsed Time: 0:00:09 ETA: 0:01:55Adaptation Done ... 8% (7 of 78) |## | Elapsed Time: 0:00:10 ETA: 0:01:55Adaptation Done ... 10% (8 of 78) |## | Elapsed Time: 0:00:12 ETA: 0:01:48Adaptation Done ... 11% (9 of 78) |## | Elapsed Time: 0:00:13 ETA: 0:01:44Adaptation Done ... 12% (10 of 78) |### | Elapsed Time: 0:00:15 ETA: 0:01:41Adaptation Done ... 14% (11 of 78) |### | Elapsed Time: 0:00:16 ETA: 0:01:35Adaptation Done ... 15% (12 of 78) |### | Elapsed Time: 0:00:18 ETA: 0:01:34Adaptation Done ... 16% (13 of 78) |#### | Elapsed Time: 0:00:19 ETA: 0:01:36Adaptation Done ... 17% (14 of 78) |#### | Elapsed Time: 0:00:21 ETA: 0:01:38Adaptation Done ... 19% (15 of 78) |#### | Elapsed Time: 0:00:22 ETA: 0:01:39Adaptation Done ... 20% (16 of 78) |#### | Elapsed Time: 0:00:24 ETA: 0:01:35Adaptation Done ... 21% (17 of 78) |##### | Elapsed Time: 0:00:25 ETA: 0:01:31Adaptation Done ... 23% (18 of 78) |##### | Elapsed Time: 0:00:27 ETA: 0:01:31Adaptation Done ... 24% (19 of 78) |##### | Elapsed Time: 0:00:28 ETA: 0:01:31Adaptation Done ... 25% (20 of 78) |###### | Elapsed Time: 0:00:30 ETA: 0:01:27Adaptation Done ... 26% (21 of 78) |###### | Elapsed Time: 0:00:31 ETA: 0:01:24Adaptation Done ... 28% (22 of 78) |###### | Elapsed Time: 0:00:33 ETA: 0:01:25Adaptation Done ... 29% (23 of 78) |####### | Elapsed Time: 0:00:34 ETA: 0:01:22Adaptation Done ... 30% (24 of 78) |####### | Elapsed Time: 0:00:36 ETA: 0:01:17Adaptation Done ... 32% (25 of 78) |####### | Elapsed Time: 0:00:37 ETA: 0:01:16Adaptation Done ... 33% (26 of 78) |######## | Elapsed Time: 0:00:39 ETA: 0:01:19Adaptation Done ... 34% (27 of 78) |######## | Elapsed Time: 0:00:40 ETA: 0:01:15Adaptation Done ... 35% (28 of 78) |######## | Elapsed Time: 0:00:42 ETA: 0:01:13Adaptation Done ... 37% (29 of 78) |######## | Elapsed Time: 0:00:43 ETA: 0:01:08Adaptation Done ... 38% (30 of 78) |######### | Elapsed Time: 0:00:45 ETA: 0:01:06Adaptation Done ... 39% (31 of 78) |######### | Elapsed Time: 0:00:46 ETA: 0:01:12Adaptation Done ... 41% (32 of 78) |######### | Elapsed Time: 0:00:48 ETA: 0:01:08Adaptation Done ... 42% (33 of 78) |########## | Elapsed Time: 0:00:49 ETA: 0:01:01Adaptation Done ... 43% (34 of 78) |########## | Elapsed Time: 0:00:50 ETA: 0:01:02Adaptation Done ... 44% (35 of 78) |########## | Elapsed Time: 0:00:52 ETA: 0:01:03Adaptation Done ... 46% (36 of 78) |########### | Elapsed Time: 0:00:53 ETA: 0:01:00Adaptation Done ... 47% (37 of 78) |########### | Elapsed Time: 0:00:55 ETA: 0:01:00Adaptation Done ... 48% (38 of 78) |########### | Elapsed Time: 0:00:56 ETA: 0:01:01Adaptation Done ... 50% (39 of 78) |############ | Elapsed Time: 0:00:58 ETA: 0:01:00Adaptation Done ... 51% (40 of 78) |############ | Elapsed Time: 0:01:00 ETA: 0:00:59Adaptation Done ... 52% (41 of 78) |############ | Elapsed Time: 0:01:01 ETA: 0:00:57Adaptation Done ... 53% (42 of 78) |############ | Elapsed Time: 0:01:03 ETA: 0:00:54Adaptation Done ... 55% (43 of 78) |############# | Elapsed Time: 0:01:04 ETA: 0:00:52Adaptation Done ... 56% (44 of 78) |############# | Elapsed Time: 0:01:06 ETA: 0:00:51Adaptation Done ... 57% (45 of 78) |############# | Elapsed Time: 0:01:07 ETA: 0:00:49Adaptation Done ... 58% (46 of 78) |############## | Elapsed Time: 0:01:09 ETA: 0:00:47Adaptation Done ... 60% (47 of 78) |############## | Elapsed Time: 0:01:10 ETA: 0:00:47Adaptation Done ... 61% (48 of 78) |############## | Elapsed Time: 0:01:12 ETA: 0:00:46Adaptation Done ... 62% (49 of 78) |############### | Elapsed Time: 0:01:13 ETA: 0:00:41Adaptation Done ... 64% (50 of 78) |############### | Elapsed Time: 0:01:14 ETA: 0:00:39Adaptation Done ... 65% (51 of 78) |############### | Elapsed Time: 0:01:16 ETA: 0:00:36Adaptation Done ... 66% (52 of 78) |################ | Elapsed Time: 0:01:17 ETA: 0:00:32Adaptation Done ... 67% (53 of 78) |################ | Elapsed Time: 0:01:18 ETA: 0:00:30Adaptation Done ... 69% (54 of 78) |################ | Elapsed Time: 0:01:19 ETA: 0:00:30Adaptation Done ... 70% (55 of 78) |################ | Elapsed Time: 0:01:21 ETA: 0:00:29Adaptation Done ... 71% (56 of 78) |################# | Elapsed Time: 0:01:22 ETA: 0:00:29Adaptation Done ... 73% (57 of 78) |################# | Elapsed Time: 0:01:24 ETA: 0:00:31Adaptation Done ... 74% (58 of 78) |################# | Elapsed Time: 0:01:25 ETA: 0:00:30Adaptation Done ... 75% (59 of 78) |################## | Elapsed Time: 0:01:27 ETA: 0:00:27Adaptation Done ... 76% (60 of 78) |################## | Elapsed Time: 0:01:28 ETA: 0:00:26Adaptation Done ... 78% (61 of 78) |################## | Elapsed Time: 0:01:30 ETA: 0:00:25Adaptation Done ... 79% (62 of 78) |################### | Elapsed Time: 0:01:31 ETA: 0:00:23Adaptation Done ... 80% (63 of 78) |################### | Elapsed Time: 0:01:33 ETA: 0:00:21Adaptation Done ... 82% (64 of 78) |################### | Elapsed Time: 0:01:34 ETA: 0:00:21Adaptation Done ... 83% (65 of 78) |#################### | Elapsed Time: 0:01:36 ETA: 0:00:19Adaptation Done ... 84% (66 of 78) |#################### | Elapsed Time: 0:01:37 ETA: 0:00:18Adaptation Done ... 85% (67 of 78) |#################### | Elapsed Time: 0:01:39 ETA: 0:00:16Adaptation Done ... 87% (68 of 78) |#################### | Elapsed Time: 0:01:40 ETA: 0:00:13Adaptation Done ... 88% (69 of 78) |##################### | Elapsed Time: 0:01:42 ETA: 0:00:13Adaptation Done ... 89% (70 of 78) |##################### | Elapsed Time: 0:01:43 ETA: 0:00:12Adaptation Done ... 91% (71 of 78) |##################### | Elapsed Time: 0:01:44 ETA: 0:00:10Adaptation Done ... 92% (72 of 78) |###################### | Elapsed Time: 0:01:46 ETA: 0:00:09Adaptation Done ... 93% (73 of 78) |###################### | Elapsed Time: 0:01:48 ETA: 0:00:07Adaptation Done ... 94% (74 of 78) |###################### | Elapsed Time: 0:01:49 ETA: 0:00:05Adaptation Done ... 96% (75 of 78) |####################### | Elapsed Time: 0:01:51 ETA: 0:00:04Adaptation Done ... 97% (76 of 78) |####################### | Elapsed Time: 0:01:52 ETA: 0:00:02Traceback (most recent call last): File "main.py", line 684, in entry_test(cfg, test_or_valid, cmd_args.model_path,cmd_args.confusion) File "main.py", line 572, in entry_test test_perf = validate(cfg.EXP.TASK, loader_test, model, cfg.EXP.DATASET, cfg.ADAPT, confusion) File "main.py", line 228, in validate model = adapt_tent(inp,model,adapt) File "main.py", line 45, in adapt_tent tent_helper.forward_and_adapt(data,model,optimizer_tent) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/autograd/grad_mode.py", line 96, in decorate_enable_grad return func(*args, **kwargs) File "~/ModelNet40-C/third_party/tent_helper.py", line 52, in forward_and_adapt outputs = model(**x) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in call result = self.forward(*input, **kwargs) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 152, in forward outputs = self.parallel_apply(replicas, inputs, kwargs) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 162, in parallel_apply return parallel_apply(replicas, inputs, kwargs, self.device_ids[:len(replicas)]) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 85, in parallel_apply output.reraise() File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/_utils.py", line 394, in reraise raise self.exc_type(msg) ValueError: Caught ValueError in replica 0 on device 0. Original Traceback (most recent call last): File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/parallel/parallel_apply.py", line 60, in _worker output = module(*input, **kwargs) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in call result = self.forward(*input, **kwargs) File "~/ModelNet40-C/models/pct.py", line 29, in forward logit = self.model(pc) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in call result = self.forward(*input, **kwargs) File "~/ModelNet40-C/PCT_Pytorch/model.py", line 69, in forward x = F.leaky_relu(self.bn6(self.linear1(x)), negative_slope=0.2) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/modules/module.py", line 532, in call result = self.forward(*input, **kwargs) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/modules/batchnorm.py", line 107, in forward exponential_average_factor, self.eps) File "~/miniconda3/envs/modelnetc/lib/python3.7/site-packages/torch/nn/functional.py", line 1666, in batch_norm raise ValueError('Expected more than 1 value per channel when training, got input size {}'.format(size)) ValueError: Expected more than 1 value per channel when training, got input size torch.Size([1, 512])

opened by dirtycomputer 2
Error occured when generating corruptions
Hi, thank you for your excellent paper and code release. I tried to generate ModelNet40-C but when I run generate_c.py, an error occured.

Traceback (most recent call last): File "data/generate_c.py", line 401, in <module> save_data(data[:,index,:], cor, sev) File "data/generate_c.py", line 360, in save_data new_data.append(MAP[corruption](data[i],severity)) TypeError: occlusion() takes 1 positional argument but 2 were given

I saw you comment some lines in generate_c.py, then can you give me some guides for 'occlusion' and 'lidar' corruptions?

# if cor in ['occlusion', 'lidar','original']: # continue

And, if I want to use this dataset for benchmark, should I download ModelNet40-C directly using ./download.sh modelnet40_c rather than generate myself?

Thank you!
opened by NahyukLEE 2
Please CITE GDANet at the end of README.

Hi authors,

We are the authors of GDANet.

Since you use our code in your repo, it is necessary for you to cite us following the other works, as shown at the end of your README.

Thanks, Mino

opened by mutianxu 1
Clean the directory structure
Static data (e.g., dataset, pretrained models) move to a folder like data.

Dynamic data (e.g., saved execution results) move to a folder like result.

Executable code (e.g., python with main) move to a folder like src.

Libraries and modules (e.g., third-party tools/repos) move to sub-directories in src.

Configuration files (looks good now).

Script tools (*.sh files) move to scripts or in the root directory.
opened by zqzqz 0

Releases(v1.0)

v1.0(Jan 29, 2022)

Source code(tar.gz)
Source code(zip)

Repo for "Benchmarking Robustness of 3D Point Cloud Recognition against Common Corruptions" https://arxiv.org/abs/2201.12296

Related tags

Overview

Benchmarking Robustness of 3D Point Cloud Recognition against Common Corruptions

ModelNet40-C

ModelNet40-C Leaderboard

Getting Started

Requirements

Install Libraries

Download Datasets Including ModelNet40-C and Pre-trained Models

New Features

Code Organization In Originial SimpleView

Running Experiments

Training and Config files

Evaluate a pretrained model

Citation

References

Comments

PGD-based adversarial training for Pointnet++

Using the code on Kitti datasets.

About implementation for Metrics

batchnorm error

Error occured when generating corruptions

Please CITE GDANet at the end of README.

Clean the directory structure

Releases(v1.0)

v1.0(Jan 29, 2022)

Owner

Jiachen Sun

This repository contains the code used for Predicting Patient Outcomes with Graph Representation Learning (https://arxiv.org/abs/2101.03940).

Official implementation of the paper Image Generators with Conditionally-Independent Pixel Synthesis https://arxiv.org/abs/2011.13775

Pytorch implementation of Each Part Matters: Local Patterns Facilitate Cross-view Geo-localization https://arxiv.org/abs/2008.11646

https://arxiv.org/abs/2102.11005

Supplementary code for the paper "Meta-Solver for Neural Ordinary Differential Equations" https://arxiv.org/abs/2103.08561

Code for paper "A Critical Assessment of State-of-the-Art in Entity Alignment" (https://arxiv.org/abs/2010.16314)

[PyTorch] Official implementation of CVPR2021 paper "PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency". https://arxiv.org/abs/2103.05465

Official Implementation for "ReStyle: A Residual-Based StyleGAN Encoder via Iterative Refinement" https://arxiv.org/abs/2104.02699

This is an official implementation of our CVPR 2021 paper "Bottom-Up Human Pose Estimation Via Disentangled Keypoint Regression" (https://arxiv.org/abs/2104.02300)

Code for the paper: Learning Adversarially Robust Representations via Worst-Case Mutual Information Maximization (https://arxiv.org/abs/2002.11798)

ISTR: End-to-End Instance Segmentation with Transformers (https://arxiv.org/abs/2105.00637)

Non-Official Pytorch implementation of "Face Identity Disentanglement via Latent Space Mapping" https://arxiv.org/abs/2005.07728 Using StyleGAN2 instead of StyleGAN

Minimal implementation of PAWS (https://arxiv.org/abs/2104.13963) in TensorFlow.

YOLO5Face: Why Reinventing a Face Detector (https://arxiv.org/abs/2105.12931)

A PyTorch implementation of EventProp [https://arxiv.org/abs/2009.08378], a method to train Spiking Neural Networks

Unofficial implementation of "Swin Transformer: Hierarchical Vision Transformer using Shifted Windows" (https://arxiv.org/abs/2103.14030)

Source code for models described in the paper "AudioCLIP: Extending CLIP to Image, Text and Audio" (https://arxiv.org/abs/2106.13043)

Unofficial implementation of Alias-Free Generative Adversarial Networks. (https://arxiv.org/abs/2106.12423) in PyTorch

source code for https://arxiv.org/abs/2005.11248 "Accelerating Antimicrobial Discovery with Controllable Deep Generative Models and Molecular Dynamics"