AgSentry: LiDAR reflectivity based obstacle detection for safety in agricultural environments

Few-shot 3D point cloud segmentation via dynamic multi-scale sparse attention with adaptive gated context enhancement

3D point clouds are widely utilized in critical vision tasks such as autonomous driving, augmented reality, and model reconstruction. Given the unstructured nature and large-scale characteristics of point cloud data, 3D point semantic segmentation often faces challenges in achieving both high accuracy and computational efficiency. To better balance accuracy and performance for domain-specific applications, this paper proposes a lightweight Rand Transformer Net (RTN), which constructs a more efficient multi-scale feature extraction module by using a random downsampling strategy and incorporates a specially designed Rand Transformer Block to capture local geometric features of point clouds. To address the issue of semantic ambiguity in boundary regions, A novel loss function, termed ABL loss, is introduced to impose additional constraints derived from labeled boundaries. Experimental results on a newly introduced Bridge Dataset, which is composed of large-scale point cloud-based bridge components, demonstrate that RTN achieves superior performance over the majority of existing point cloud semantic segmentation methods, with strong scalability and efficiency advantages in handling large-scale point cloud scenes.

Achieving rapid and high-precision 3D reconstructions of urban scenes is crucial for building the digital Earth. The key challenge lies in accurately recovering fine structural edges, but sparse point clouds provide limited geometric constraints. Although dense point clouds increase reconstruction accuracy, their high computational cost restricts their practical use. Moreover, most existing methods overlook line features as essential geometric primitives for surface modeling, reducing both accuracy and efficiency. To address these challenges, a surface reconstruction method that integrates 3D line features and sparse point clouds based on multiview images is proposed. First, a line feature sampling method based on point cloud density, which combines line features with sparse point clouds in 3D space, is proposed. Second, accurate edge visual information of the surface model is obtained by constructing the average ray of the line features. Third, an energy function with a line feature constraint is proposed to ensure the integrity of the surface edge. Quantitative evaluations indicate that the median surface-to-point distance decreases by 15.3% in small-scale scenes and 14.9% in large-scale scenes. The reconstruction time is reduced by 98.3% compared with that of dense matching, indicating the high accuracy and efficiency of the proposed approach.

NeuralBoneReg: An instance-specific label-free point cloud-based method for multi-modal bone surface registration.

L2G-Net: Local-to-global feature enhancement via cluster tokens for 3D place recognition.

HyPCV-Former: Hyperbolic spatio-temporal transformer for 3D point cloud video anomaly detection

Multi-view 3D point cloud registration method based on generated multi-scale information granules

Structural health monitoring of underground tunnels increasingly uses advanced sensing and data-driven methods. Laser-scanned 3D point clouds capture spatially rich measurements of segmental tunnel linings and require segmentation as a prerequisite for downstream analysis. Deep learning (DL) is effective for point-cloud segmentation, but scarce datasets and costly annotation limit practical use. This paper presents Tunnel Scanner , a high-fidelity simulator that synthesises realistic tunnel point clouds with automatic annotation. The plug-and-play module Hybrid Position–Normal Local Spatial Encoding embeds geometric priors into DL backbones and combines with transfer learning (TL) to exploit synthetic data for domain adaptation. Models trained only on synthetic data achieved 71.9% mean Intersection-over-Union (mIoU) and 86.6% Overall Accuracy (OA), and TL increased performance to at least 78.8% mIoU and 90.9% OA with limited real data. This paper highlights geometry-informed data synthesis as a viable augmentation approach for digital inspection and asset management of large-scale tunnels. • Develop a high-fidelity simulator to address the scarcity of tunnel point clouds. • Propose a plug-and-play module encoding geometric features into DL backbones. • Investigate transfer learning to enhance 3D Sim-to-Real domain adaptation. • The end-to-end framework yields ＋24% segmentation accuracy with limited real data.

3D point cloud segmentation based on updated restrictions for contact and intersection objects

Multimodal fusion of 3D point cloud and intraoperative imaging to enhance surgical robot navigation

Point-DPA: Unifying contrastive and generative learning for 3D point cloud understanding via dynamic prototypes

GCG-PROSAC: a geometric consistency-based robust estimation algorithm for accelerated 3D point cloud registration

GRV: Adversarial defense for 3D point clouds using geometric restoration and multi-model voting

Estimating the 6-DoF posture of parts in assembly-based modeling is a critical task in the fields of computer graphics, computer vision and robotics. A typical scenario involves enabling a machine agent to automatically assemble IKEA furniture using the provided parts. This paper presents HiFormer, a novel Hierarchical Transformer with Box-packed Positional Encoding, designed for highly automatic 3D part assembly. Our method addresses three important issues commonly encountered in 3D part assembly: 1) How to mitigate the overfitting problem associated with Transformer-based feature learning for 3D point clouds? 2) How to effectively model the relationships between the intragroup and intergroup parts? 3) How to compute positional encoding and integrate it into the Transformer for parts with diverse geometric forms in the coarse-to-fine assembly process? These challenges are tackled through three key contributions: 1) a multi-task 3D Swin Transformer with a two-stage training strategy for feature extraction, 2) a novel hierarchical Transformer for capturing part relationships at flattening, intragroup, and intergroup levels, and 3) an innovative box-packed positional encoding that enhances the Transformer by incorporating query, key, and value information derived from relative box positions. On the PartNet benchmark, our method outperforms the state-of-the-art PWH-MP model on three representative categories-Chair, Table, and Lamp-, achieving average improvements of 2.84% in Part Accuracy (PA) and 3.72% in Connection Accuracy (CA) for diversity modeling (with noise), and 3.55% in PA and 3.21% in CA for deterministic modeling (without noise).

To evaluate an automated computed tomography (CT)-derived three-dimensional (3D) model-based planning pipeline for C1-C2 screw fixation that generates multiple anatomically grounded trajectories per side, and to test the hypothesis that redundancy-by-design provides at least one Grade 0-1 candidate per evaluable side. This retrospective, single-center study with supportive external dataset testing on VerSe 2019 used CT-derived 3D surface models, PointNet++ (aligned) point cloud segmentation, and geometric landmarking to generate four candidate screw axes for C1 and three for C2 per side. Internal and external candidates were graded independently by two spine-surgeon raters using the Gertzbein-Robbins scale. Conservative consensus was defined as Grade 0-1 only when both raters graded a candidate as Grade 0 or 1. The primary endpoint was side-level coverage of at least one Grade 0-1 candidate; secondary endpoints included candidate-level Grade 0-1 rate, segmentation performance mean intersection-over-union (mIoU), landmark localization error, and inter-rater agreement. On the internal test set, segmentation achieved a mean region-wise mIoU of 88.8%, with landmark localization errors of approximately 1mm. By conservative consensus, 93.6% of internal candidates were Grade 0-1 (96.1% and 93.8% by individual raters). On the external VerSe 2019 dataset, the conservative-consensus Grade 0-1 rate was 88.5% overall, including 82.1% for C1 and 97.0% for C2. Every evaluable internal and external side had at least one Grade 0-1 candidate. Inter-rater agreement was substantial in both datasets. The CT-derived 3D point-cloud and geometric planning pipeline generated multiple C1-C2 screw trajectory candidates and the redundancy-by-design strategy provides retrospective preclinical feasibility evidence for a reproducible preoperative decision-support framework. Further multicentre, vascular-risk-aware, cadaveric/intraoperative, and prospective validation is required before clinical deployment.

An integrated framework for surface reconstruction from incomplete point clouds based on constraint feedforward migration.

Based on the complementary and enhanced fusion of 3D point clouds and 2D RGB images, this paper designs an end-to-end learning framework-Point Cloud Enhanced Depth Pixel Fusion Network (PEPF-Net), aimed at enabling robots to achieve accurate 3D perception of unstructured environments. In the process, we address four key problems in 3D perception tasks: enhancing RGB representation using the reflection intensity and depth information of point clouds to generate Depth-RGB Pixel (D-Pixel); proposing Point-by-Point Vector Attention (PVA-Net) to model the vector relationships of point clouds, to obtain deep-level point cloud features, and to achieve direct and effective fusion of heterogeneous data; designing a Layered-Transformer (L-TsfmNet) feature extractor to hierarchically extract D-Pixel features; proposing Variable Window Self-attention (VS-a) to focus on the relationships between local "window tokens" and avoid the complexity of global computation. Extensive experiments on the KITTI dataset demonstrate that PEPF-Net outperforms the currently common advanced environmental 3D perception algorithms.

An efficient 3D point clouds outlier noise removal method based on feature consistency

To evaluate the accuracy of a high-precision navigation system that projects a three-dimensional (3D) kidney model onto surgical images by limiting its application to moments of minimal organ deformation during partial nephrectomy (PN). We analyzed 29 patients who underwent PN at Kyoto University Hospital and Kobe City Medical Center General Hospital in Japan. 3D models of the kidney and tumor were generated using DICOM data, whereas 3D point clouds of the surgical field were obtained using stereo camera recordings. Noise reduction processing was applied to the camera-derived point clouds. Registration between the computed tomography-derived models and camera-derived point clouds was performed using the closest iterative point, and the accuracy was assessed using the root mean squared error. We evaluated the effects of the point-cloud surface area and camera-to-target distance on the registration accuracy. Without noise reduction, the median registration error was 2.33 mm, whereas noise reduction improved the accuracy by 1.83 mm. The accuracy was significantly higher when the camera-to-target distance was shorter, with and without noise reduction. The surface area was inversely correlated with the accuracy without noise reduction, but no significant correlation was observed with noise reduction. Focusing on moments with minimal organ deformation, we demonstrated that high-precision surgical navigation is achievable in PN using actual surgical recordings. This may contribute to improved tumor localization and the preservation of renal function.