Geometric Adversarial Attacks and Defenses on 3D Point Clouds

Lang, Itai; Kotlicki, Uriel; Avidan, Shai

doi:10.48550/arxiv.2012.05657

Cited by 3 publications

(3 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…[36] presents an in-depth study showing how adversarial training behaves in point cloud classification. However, existing works only focus on improving the model's robustness to perturbations of random point shifting or removing [12,16,19,43,45,52].…”

Section: Adversarial Trainingmentioning

confidence: 99%

ART-Point: Improving Rotation Robustness of Point Cloud Classifiers via Adversarial Rotation

Wang¹,

Yang²,

Tao³

2022

Preprint

View full text Add to dashboard Cite

Point cloud classifiers with rotation robustness have been widely discussed in the 3D deep learning community. Most proposed methods either use rotation invariant descriptors as inputs or try to design rotation equivariant networks. However, robust models generated by these methods have limited performance under clean aligned datasets due to modifications on the original classifiers or input space. In this study, for the first time, we show that the rotation robustness of point cloud classifiers can also be acquired via adversarial training with better performance on both rotated and clean datasets. Specifically, our proposed framework named ART-Point regards the rotation of the point cloud as an attack and improves rotation robustness by training the classifier on inputs with Adversarial RoTations. We contribute an axis-wise rotation attack that uses back-propagated gradients of the pre-trained model to effectively find the adversarial rotations. To avoid model over-fitting on adversarial inputs, we construct rotation pools that leverage the transferability of adversarial rotations among samples to increase the diversity of training data. Moreover, we propose a fast one-step optimization to efficiently reach the final robust model. Experiments show that our proposed rotation attack achieves a high success rate and ART-Point can be used on most existing classifiers to improve the rotation robustness while showing better performance on clean datasets than state-of-the-art methods.

show abstract

Section: Adversarial Trainingmentioning

confidence: 99%

ART-Point: Improving Rotation Robustness of Point Cloud Classifiers via Adversarial Rotation

Wang¹,

Yang²,

Tao³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Task Description: In this task, we aim to generate realistic adversarial traffic scenes against point cloud segmentation algorithms, while satisfying certain semantic knowledge rules. To generated adversarial LiDAR scenes containing various fore-/background rather than the point cloud of single 3D object as existing works Lang et al [2020], Sun et al [2020a], a couple of challenges should be considered: First, LiDAR scenes with millions of points are hard to be directly operated; Second, generated scenes need to be realistic and follow traffic rules. Since there are no existing baselines to directly compare with, we implement three methods: (1) Point Attack: a point-wise attack baseline Xiang et al [2019] that adds small disturbance to points; (2) Pose Attack: a scene generation method developed by us that searches poses of a fixed number of vehicles; (3) Scene Attack: a semantically controllable traffic generative method based on our T-VAE and SCG.…”

Section: Adversarial Traffic Scenes Generationmentioning

confidence: 99%

Semantically Adversarial Driving Scenario Generation with Explicit Knowledge Integration

Ding¹,

Hu²,

Li³

et al. 2021

Preprint

View full text Add to dashboard Cite

Deep Generative Models (DGMs) are known for their superior capability in generating realistic data. Extending purely data-driven approaches, recent specialized DGMs may satisfy additional controllable requirements such as embedding a traffic sign in a driving scene, by manipulating patterns implicitly in the neuron or feature level. In this paper, we introduce a novel method to incorporate domain knowledge explicitly in the generation process to achieve semantically controllable scene generation. We categorize our knowledge into two types to be consistent with the composition of natural scenes, where the first type represents the property of objects and the second type represents the relationship among objects. We then propose a tree-structured generative model to learn complex scene representation, whose nodes and edges are naturally corresponding to the two types of knowledge respectively. Knowledge can be explicitly integrated to enable semantically controllable scene generation by imposing semantic rules on properties of nodes and edges in the tree structure. We construct a synthetic example to illustrate the controllability and explainability of our method in a clean setting. We further extend the synthetic example to realistic autonomous vehicle driving environments and conduct extensive experiments to show that our method efficiently identifies adversarial traffic scenes against different state-of-the-art 3D point cloud segmentation models satisfying the traffic rules specified as the explicit knowledge.Preprint. Under review.

show abstract

“…Denoising. Many deep learning techniques have been presented for point cloud denoising [Hermosilla et al 2019;Luo and Hu 2020;Lang et al 2020]. Indeed, using the denoised cloud requires estimating normals based on the new point locations, necessitating normal orientation.…”

Section: Neural Point Cloud Generationmentioning

confidence: 99%

Orienting point clouds with dipole propagation

et al. 2021

View full text Add to dashboard Cite

Establishing a consistent normal orientation for point clouds is a notoriously difficult problem in geometry processing, requiring attention to both local and global shape characteristics. The normal direction of a point is a function of the local surface neighborhood; yet, point clouds do not disclose the full underlying surface structure. Even assuming known geodesic proximity, calculating a consistent normal orientation requires the global context. In this work, we introduce a novel approach for establishing a globally consistent normal orientation for point clouds. Our solution separates the local and global components into two different sub-problems. In the local phase, we train a neural network to learn a coherent normal direction per patch ( i.e. , consistently oriented normals within a single patch). In the global phase, we propagate the orientation across all coherent patches using a dipole propagation. Our dipole propagation decides to orient each patch using the electric field defined by all previously orientated patches. This gives rise to a global propagation that is stable, as well as being robust to nearby surfaces, holes, sharp features and noise.

show abstract

Geometric Adversarial Attacks and Defenses on 3D Point Clouds

Cited by 3 publications

References 39 publications

ART-Point: Improving Rotation Robustness of Point Cloud Classifiers via Adversarial Rotation

ART-Point: Improving Rotation Robustness of Point Cloud Classifiers via Adversarial Rotation

Semantically Adversarial Driving Scenario Generation with Explicit Knowledge Integration

Orienting point clouds with dipole propagation

Contact Info

Product

Resources

About