Explore the vast range of titles available.

Existing 3D point cloud enhancement methods typically rely on artificially designed geometric transformations or local blending strategies, which are prone to introducing illogical deformations, struggle to preserve global structure, and exhibit insufficient adaptability to diverse degradation patterns. To address these limitations, this paper proposes SFD-ADNet—an adaptive deformation framework based on a dual spatial–frequency domain. It achieves 3D point cloud augmentation by explicitly learning deformation parameters rather than applying predefined perturbations. By jointly modeling spatial structural dependencies and spectral features, SFD-ADNet generates augmented samples that are both structurally aware and task-relevant. In the spatial domain, a hierarchical sequence encoder coupled with a bidirectional Mamba-based deformation predictor captures long-range geometric dependencies and local structural variations, enabling adaptive position-aware deformation control. In the frequency domain, a multi-scale dual-channel mechanism based on adaptive Chebyshev polynomials separates low-frequency structural components from high-frequency details, allowing the model to suppress noise-sensitive distortions while preserving the global geometric skeleton. The two deformation predictions dynamically fuse to balance structural fidelity and sample diversity. Extensive experiments conducted on ModelNet40-C and ScanObjectNN-C involved synthetic CAD models and real-world scanned point clouds under diverse perturbation conditions. SFD-ADNet, as a universal augmentation module, reduces the mCE metrics of PointNet++ and different backbone networks by over 20%. Experiments demonstrate that SFD-ADNet achieves state-of-the-art robustness while preserving critical geometric structures. Furthermore, models enhanced by SFD-ADNet demonstrate consistently improved robustness against diverse point cloud attacks, validating the efficacy of adaptive space-frequency deformation in robust point cloud learning.

Recent years have witnessed significant advancements in deep learning-based 3D object detection, leading to its widespread adoption in numerous applications. As 3D object detectors become increasingly crucial for security-critical tasks, it is imperative to understand their robustness against adversarial attacks. This paper presents the first comprehensive evaluation and analysis of the robustness of LiDAR-based 3D detectors under adversarial attacks. Specifically, we extend three distinct adversarial attacks to the 3D object detection task, benchmarking the robustness of state-of-the-art LiDAR-based 3D object detectors against attacks on the KITTI and Waymo datasets. We further analyze the relationship between robustness and detector properties. Additionally, we explore the transferability of cross-model, cross-task, and cross-data attacks. Thorough experiments on defensive strategies for 3D detectors are conducted, demonstrating that simple transformations like flipping provide little help in improving robustness when the applied transformation strategy is exposed to attackers. Finally, we propose balanced adversarial focal training, based on conventional adversarial training, to strike a balance between accuracy and robustness. Our findings will facilitate investigations into understanding and defending against adversarial attacks on LiDAR-based 3D object detectors, thus advancing the field. The source code is publicly available at https://github.com/Eaphan/Robust3DOD.

3D point cloud models are widely applied in safety-critical scenes, which delivers an urgent need to obtain more solid proofs to verify the robustness of models. Existing verification method for point cloud model is time-expensive and computationally unattainable on large networks. Additionally, they cannot handle the complete PointNet model with joint alignment network that contains multiplication layers, which effectively boosts the performance of 3D models. This motivates us to design a more efficient and general framework to verify various architectures of point cloud models. The key challenges in verifying the large-scale complete PointNet models are addressed as dealing with the cross-non-linearity operations in the multiplication layers and the high computational complexity of high-dimensional point cloud inputs and added layers. Thus, we propose an efficient verification framework, 3DVerifier, to tackle both challenges by adopting a linear relaxation function to bound the multiplication layer and combining forward and backward propagation to compute the certified bounds of the outputs of the point cloud models. Our comprehensive experiments demonstrate that 3DVerifier outperforms existing verification algorithms for 3D models in terms of both efficiency and accuracy. Notably, our approach achieves an orders-of-magnitude improvement in verification efficiency for the large network, and the obtained certified bounds are also significantly tighter than the state-of-the-art verifiers. We release our tool 3DVerifier via https://github.com/TrustAI/3DVerifier for use by the community.

Standard registration algorithms need to be independently applied to each surface to register, following careful pre-processing and hand-tuning. Recently, learning-based approaches have emerged that reduce the registration of new scans to running inference with a previously-trained model. The potential benefits are multifold: inference is typically orders of magnitude faster than solving a new instance of a difficult optimization problem, deep learning models can be made robust to noise and corruption, and the trained model may be re-used for other tasks, e.g. through transfer learning. In this paper, we cast the registration task as a surface-to-surface translation problem, and design a model to reliably capture the latent geometric information directly from raw 3D face scans. We introduce Shape-My-Face (SMF), a powerful encoder-decoder architecture based on an improved point cloud encoder, a novel visual attention mechanism, graph convolutional decoders with skip connections, and a specialized mouth model that we smoothly integrate with the mesh convolutions. Compared to the previous state-of-the-art learning algorithms for non-rigid registration of face scans, SMF only requires the raw data to be rigidly aligned (with scaling) with a pre-defined face template. Additionally, our model provides topologically-sound meshes with minimal supervision, offers faster training time, has orders of magnitude fewer trainable parameters, is more robust to noise, and can generalize to previously unseen datasets. We extensively evaluate the quality of our registrations on diverse data. We demonstrate the robustness and generalizability of our model with in-the-wild face scans across different modalities, sensor types, and resolutions. Finally, we show that, by learning to register scans, SMF produces a hybrid linear and non-linear morphable model. Manipulation of the latent space of SMF allows for shape generation, and morphing applications such as expression transfer in-the-wild. We train SMF on a dataset of human faces comprising 9 large-scale databases on commodity hardware.

Most of the current simultaneous localization and mapping (SLAM) algorithms are realized based on the assumption of a static environment. However, most environments are dynamic in the real world. Therefore, we propose a semantic SLAM algorithm for a dynamic environment. Specifically, we obtain the semantic information through deep learning, combine it with the position information of point cloud to filter out dynamic objects and alleviate the accuracy degradation caused by dynamic obstacles. Due to factors such as noise and occlusion, there are some outliers present within the point cloud data, and these data points can adversely impact the performance of the algorithm. To solve this, we utilize semantic information consistency to adjust the weighting of the error function, thereby mitigating the impact of outliers and enhancing the robustness of the front‐end odometry. On this basis, we further propose a multi‐frame verification method based on descriptors to optimize the back‐end loop closure detection algorithm. The experimental results show that our method enhances accuracy and robustness compared with the benchmark.

Discontinuities largely influence the mechanical properties of rock joints. However, traditional discontinuity recognition methods often require manual intervention during processing. This paper proposes a new deep-learning-based method for discontinuity recognition using 3D point clouds. A neighborhood PCA-weighted oriented contraction method is proposed to extract point cloud skeletons as discontinuity intersection lines. Then an optimal color mapping (OCM) method is proposed to establish the optimal mapping relationship between 3D normal vectors and RGB, converting 3D point clouds to 2D OCM images for discontinuity segmentation. The convolutional neural network of Mask R-CNN is adopted to segment discontinuities from OCM images. Finally, 3D discontinuities can be generated from discontinuity-segmented OCM images. Forty-two rock slope image sequences and a rock slope point cloud are collected and labeled, generating 4632 OCM images including 430,613 discontinuity planes after data augmentation for training. Three cases of rock slopes and rock tunnel excavation faces are adopted for testing. The average recognition time per 3D point cloud model is approximately 12 s due to the high recognition efficiency of Mask R-CNN for 2D images. The results show the proposed method can recognize discontinuities close to manual judgements with high accuracy, good robustness to point cloud density variations, and good adaptability to different rock engineering scenarios.HighlightsAn NPW-OC method is proposed to extract point cloud skeletons.An OCM method is proposed to assign 3D normal vectors with optimal RGB.OCM images are generated to assign discontinuities with different and uniform color.Deep-learning-based method is used for intelligent recognition of discontinuities.Conversion of discontinuity recognition from 3D point clouds to 2D OCM images.The results show good efficiency, accuracy, robustness, and adaptability .

Accurate cloud identification is crucial for understanding the rapid evolution of weather systems, improving the accuracy of short-term forecasts, and ensuring aviation safety. Compared with traditional cloud image recognition methods, deep learning technology has advantages such as automatic learning of complex features, high-precision recognition, and strong robustness in changing environments, providing more reliable and detailed cloud information. This study utilized 256 cloud image observation data points collected by an all-sky imager from 3 to 30 November 2023, at the Tunchang County Meteorological Bureau in Hainan Province (19°21′N, 110°06′ E). A Convolutional Neural Network (CNN) model was employed for cloud image recognition. The results show that in terms of cloud recognition, the constructed CNN model achieved an accuracy rate, recall rate, and F1 score of 100%, 91%, and 95%, respectively, for clear skies and stratus clouds, cumulus clouds, and cirrus clouds, with an average recognition accuracy rate of 95%. In terms of cloud cover detection, when comparing the Normalized Red Blue Ratio (NRBR) and K-Means clustering algorithm with the system’s built-in monitoring results, the NRBR method performed optimally in cloud region segmentation, with cloud cover estimates closer to the actual distribution. In summary, deep learning technology demonstrates higher accuracy and strong robustness in all-sky cloud image recognition.

LiDAR sensors have been shown to generate data with various common corruptions, which seriously affect their applications in 3D vision tasks, particularly object detection. At the same time, it has been demonstrated that traditional defense strategies, including adversarial training, are prone to suffering from gradient confusion during training. Moreover, they can only improve their robustness against specific types of data corruption. In this work, we propose LiDARPure, which leverages the powerful generation ability of diffusion models to purify corruption in the LiDAR scene data. By dividing the entire scene into voxels to facilitate the processes of diffusion and reverse diffusion, LiDARPure overcomes challenges induced from adversarial training, such as sparse point clouds in large-scale LiDAR data and gradient confusion. In addition, we utilize the latent geometric features of a scene as a condition to assist the generation of diffusion models. Detailed experiments show that LiDARPure can effectively purify 19 common types of LiDAR data corruption. Further evaluation results demonstrate that it can improve the average precision of 3D object detectors to an extent of 20% in the face of data corruption, much higher than existing defence strategies.

Autonomous navigation in large-scale and complex environments in the absence of a GPS signal is a fundamental challenge encountered in a variety of applications. Since 3-D scans provide inherent robustness to ambient illumination changes and the type of the surface texture, we present Point Cloud Map-based Navigation (PCMN), a robust robot navigation system, based exclusively on 3-D point cloud registration between an acquired observation and a stored reference map. It provides a drift-free navigation solution, equipped with a failed registration detection capability. The backbone of the navigation system is a robust point cloud registration method, of the acquired observation to the stored reference map. The proposed registration algorithm follows a hypotheses generation and evaluation paradigm, where multiple statistically independent hypotheses are generated from local neighborhoods of putative matching points. Then, hypotheses are evaluated using a multiple consensus analysis that integrates evaluation of the point cloud feature correlation and a consensus test on the Special Euclidean Group SE(3) based on independent hypothesized estimates. The proposed PCMN is shown to achieve significantly better performance than state-of-the-art methods, both in terms of place recognition recall and localization accuracy, achieving submesh resolution accuracy, both for indoor and outdoor settings.

This study proposes an innovative deep learning algorithm for pose estimation based on point clouds, aimed at addressing the challenges of pose estimation for objects affected by the environment. Previous research on using deep learning for pose estimation has primarily been conducted using RGB-D data. This paper introduces an algorithm that utilizes point cloud data for deep learning-based pose computation. The algorithm builds upon previous work by integrating PointNet +  + technology and the classical Point Pair Features algorithm, achieving accurate pose estimation for objects across different scene scales. Additionally, an adaptive parameter-density clustering method suitable for point clouds is introduced, effectively segmenting clusters in varying point cloud density environments. This resolves the complex issue of parameter determination for density clustering in different point cloud environments and enhances the robustness of clustering. Furthermore, the LineMod dataset is transformed into a point cloud dataset, and experiments are conducted on the transformed dataset to achieve promising results with our algorithm. Finally, experiments under both strong and weak lighting conditions demonstrate the algorithm's robustness.