This repository provides the official PyTorch implementation of the following conference paper:
Eidos: Efficient, Imperceptible Adversarial 3D Point Clouds
a. To setup a conda environment:
conda env create -f environment.ymlb. We use the same model from paper Shape-invariant 3D Adversarial Point Clouds (CVPR 2022), download checkpoints from https://drive.google.com/file/d/1L25i0l6L_b1Vw504WQR8-Z0oh2FJA0G9/view and set it properly in ./checkpoint/:
[2025.5.10] Mini-batch attack added.
c. start eidos attack:
conda activate eidos_attack
python main.py --transfer_attack_method ifgm_bp_ours --task_name adv_result --exponential_steppython main.py --transfer_attack_method ifgm_bp_ours --task_name adv_result --exponential_step --surrogate_model pointnet_clspython main.py --transfer_attack_method ifgm_ours --task_name adv_result --bp_version bp2 --l2_weight 1.0 --hd_weight 0.0 --curve_weight 0.0 --cd_weight 0.0 --exponential_step --surrogate_model pointnet_clspython main.py --transfer_attack_method gsda_bp --task_name adv_result --surrogate_model pointnet_clspython main.py --transfer_attack_method ifgm_bp --task_name adv_result --surrogate_model pointnet_clspython main.py --transfer_attack_method ifgm_bp_ours_query --task_name adv_result --surrogate_model dgcnn --target_model paconv --step_size 0.32a. following the instruction at https://github.com/lulutang0608/Point-BERT to install CPP extensions
b. download PointTransformer_ModelNet1024points.pth from https://cloud.tsinghua.edu.cn/f/9be5d9dcbaeb48adb360/?dl=1 and make sure it is in checkpoint/ModelNet40/PointTransformer_ModelNet1024points.pth
python main.py --transfer_attack_method ifgm_bp_ours --dataset ModelNet40Full --data-path data/modelnet40_normal_resampled --surrogate_model pointllm_bert --task_name adv_result --exponential_step