Generating 3D Adversarial Point Clouds

By Chong Xiang, Charles Ruizhongtai Qi, Bo Li. Computer Vision and Pattern Recognition, 2018

386 citations recorded by the source corpus.

Deep neural networks are known to be vulnerable to adversarial examples which are carefully crafted instances to cause the models to make wrong predictions. While adversarial examples for 2D images and CNNs have been extensively studied, less attention has been paid to 3D data such as point clouds. Given many safety-critical 3D applications such as autonomous driving, it is important to study how adversarial point clouds could affect current deep 3D models. In this work, we propose several novel algorithms to craft adversarial point clouds against PointNet, a widely used deep neural network for point cloud processing. Our algorithms work in two ways: adversarial point perturbation and adversarial point generation. For point perturbation, we shift existing points negligibly. For point generation, we generate either a set of independent and scattered points or a small number (1-3) of point clusters with meaningful shapes such as balls and airplanes which could be hidden in the human psyche. In addition, we formulate six perturbation measurement metrics tailored to the attacks in point clouds and conduct extensive experiments to evaluate the proposed algorithms on the ModelNet40 3D shape classification dataset. Overall, our attack algorithms achieve a success rate higher than 99% for all targeted attacks.

Subfields: Artificial Intelligence, Materials Chemistry

View Paper: https://www.semanticscholar.org/paper/c783584140a727bf3951c0489a857750e975d3ad

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Materials Chemistry

Citations in the corpus

Adversarial Sensor Attack on LiDAR-based Perception in Autonomous Driving (2019, background): “An ongoing work [49] studies the generation of 3D adversarial point clouds.”
Adversarial Attacks and Defenses in Deep Learning (2020, background): “[44], the authors ﬁrst extend the C&W attack to the 3D point-cloud DL models.”
Physically Realizable Adversarial Examples for LiDAR Object Detection (2020, background): “Several previous works have shown successful attacks [36, 39, 43] with point cloud perturbations, generating salient modiﬁcations by adding, re-Vehicle In this work we produce a physically realizable adversarial object that can make vehicles “invisible”.”
Towards Robust LiDAR-based Perception in Autonomous Driving: General Black-box Adversarial Sensor Attack and Countermeasures (2020, background): “Point cloud classiﬁcation models have been demonstrated vulnerable to adversarial perturbations [57, 60, 64]”
Review: deep learning on 3D point clouds (2020, methodology): “The global maxpooling applied select the feature vector in a ”winner–take –all” [26] principle, making it very susceptible to targetted adversarial attack as demonstrated in [27].”
Advances in adversarial attacks and defenses in computer vision: A survey (2021, background): “Other recent examples of attacks on point clouds and 3D data (and their defenses) include [181], [182], [183], [184], [185].”