Search Results (7,082)

Large-scale terrestrial laser scanning (TLS) point clouds are increasingly used for applications such as digital twins and cultural heritage documentation; however, removing unwanted human points captured during acquisition remains a largely manual and time-consuming process. This study proposes a geometry-aware framework for automatically removing human noise from TLS point clouds by projecting 2D instance segmentation masks (obtained using You Only Look Once (YOLO) v8 with an instance segmentation head) into 3D space and validating candidates through multi-stage geometric filtering. To suppress false positives induced by reprojection misalignment and planar background structures (e.g., walls and ground), we introduce projection-followed geometric validation (or “geometric gating”) using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and principal component analysis (PCA)-based planarity analysis, followed by cluster-level plausibility checks. Experiments were conducted on two real-world outdoor TLS datasets—(i) Osaka Metropolitan University Sugimoto Campus (OMU) (82 scenes) and (ii) Jinaimachi historic district in Tondabayashi (JM) (68 scenes). The results demonstrate that the proposed method achieves high noise removal accuracy, obtaining precision/recall/intersection over union (IoU) of 0.9502/0.9014/0.8607 on OMU and 0.8912/0.9028/0.8132 on JM. Additional experiments on mobile mapping system (MMS) data from the Waymo Open Dataset demonstrate stable performance without parameter recalibration. Furthermore, quantitative and qualitative comparisons with representative time-series geometric dynamic object removal methods, including DUFOMap and BeautyMap, show that the proposed approach maintains competitive recall under a human-only ground-truth definition while reducing over-removal of static structures in TLS scenes, particularly when humans are observed in only one or a few scans due to limited revisit frequency. The end-to-end processing time with YOLOv8 was 935.62 s for 82 scenes (11.4 s/scene) on OMU and 571.58 s for 68 scenes (8.4 s/scene) on JM, supporting practical efficiency on high-resolution TLS imagery. Ablation studies further clarify the role of each stage and indicate stable performance under the observed reprojection errors. The annotated human point cloud dataset used in this study has been publicly released to facilitate reproducibility and further research on human noise removal in large-scale TLS scenes. Full article

Background: Physical inactivity remains a major public health challenge, particularly for underserved populations lacking exercise facility access. AI-powered smartphone applications with real-time human pose estimation offer scalable solutions, but they lack rigorous clinical validation. Objective: This study validates the clinical efficacy of a 16-week on-device AI-driven resistance training program using MediaPipe pose estimation technology in young adults with limited facility access. Primary outcomes included muscular strength (1RM squat), body composition, functional movement (FMS), and cardiorespiratory fitness (VO₂max). Methods: A single-group pre–post study enrolled 216 participants (mean age 23.77 ± 4.02 years; 69.2% male), with 146 (67.6%) completing the protocol. Participants performed three 30 min weekly sessions of seven compound exercises delivered via a smartphone app providing real-time pose analysis (97.2% key point accuracy, 28.6 ms inference), multimodal feedback, and personalized progression using self-selected equipment. Results: Significant improvements across all domains: muscular strength (+4.39 kg 1RM squat, p < 0.001, d = 1.148), body fat (−2.92%, p < 0.001, d = −1.373), skeletal muscle mass (+2.19 kg, p < 0.001, d = 1.433), FMS (+0.29 points, p = 0.001, d = 0.285), and VO₂max (+1.82 mL/kg/min, p < 0.001, d = 0.917). Pose classification accuracy reached 95.8% vs. physiotherapist assessment (ICC = 0.94). Conclusions: This study provides the first clinical evidence that on-device AI pose estimation enables facility-independent resistance training with outcomes comparable to traditional programs. Unlike cloud-based systems, our lightweight model (28.6 ms inference) supports real-time mobile deployment, advancing accessible precision exercise medicine. Limitations include a single-arm design and gender imbalance, warranting future RCTs with diverse cohorts. Full article

Pepper (Capsicum annuum) is a globally significant horticultural crop cultivated for its culinary, medicinal, and economic value. Traditional approaches for boosting the agricultural production of pepper, notably, expanding farmland, have become increasingly unsustainable. Recent advancements in artificial intelligence and 3D computer vision have started to transform crop cultivation and phenotyping, which has shed new light on increasing production by advanced breeding. However, currently, the field still lacks 3D pepper data that contains enough detail for organ-level analysis. Therefore, we propose Pepper-4D, a new, high-precision 4D point cloud dataset that records both the spatial structure and temporal development of pepper plants across various continuous growth stages. Our dataset is divided into three subsets, including a total of 916 individual point clouds from 29 indoor-cultivated pepper plant samples. Our dataset provides manual annotations at both the plant-level and organ-level, supporting phenotyping tasks such as pepper growth status classification, organ semantic segmentation, organ instance segmentation, organ growth tracking, new organ detection, and even the generation of synthetic 3D pepper plants. Full article

Accurate non-contact individual identification of pigs is crucial for their intelligent and efficient management. However, traditional recognition technologies generally suffer from weak local feature expression, feature redundancy, and insufficient channel importance modeling. To address these challenges, this study proposes a novel network model, DK-EffiPointMLP, for individual identification based on 3D dorsal point clouds. The model integrates a Dual-branch Local Feature enhancement module (DLF) and an Efficient Partial Convolution-Residual Refinement module (EffiConv). Specifically, the DLF module adopts a dual-branch structure of KNN and dilated KNN to expand the receptive field, while the EffiConv module combines 1D convolution with the SE mechanism to strengthen key channel modeling. To evaluate the model, a dataset of 10 individual pigs with 8411 samples was constructed. Experimental results show that DK-EffiPointMLP achieves accuracies of 96.86% on this self-built dataset and 95.2% on ModelNet40. When re-training all baseline models under the same pipeline and preprocessing protocols, our model outperformed existing mainstream models by 2.74 and 1.1 percentage points, respectively. This approach provides an efficient solution for automated management in commercial farming. Full article

The Geological Strength Index (GSI) is a crucial tool for assessing jointed rock masses, but it is often hindered by subjectivity in visual assessments. In this study, we propose a novel quantitative GSI method wherein 3D laser-scanning point clouds are used to quantitatively derive empirical rock mass indices (SR and SCR) to estimate mechanical parameters. By integrating the GSI with the Rock Block Index (RBI) and joint spacing, a framework for quantifying the Structural Rating (SR) is established. Furthermore, the Analytic Hierarchy Process (AHP) is employed to assign weights to Surface Condition Rating (SCR) factors. The results indicate that infilling materials have the most significant impact on SCR (weight 0.6334), followed by weathering (0.2605) and roughness (0.1061). This method was applied to evaluate rock masses at depths of −915 to −960 m in the Sanshandao Gold Mine. The GSI values calculated for the foot wall, ore body, and hanging wall were 38.5, 33.8, and 37.8, respectively. Validation against conventional quantitative methods demonstrated high accuracy, with a maximum relative GSI difference of 1.5 and a deformation modulus difference of only 0.227 GPa. This data-driven approach effectively reduces subjectivity and provides a reliable tool for automated geotechnical parameter estimation. Full article

The monitoring and conservation of built heritage is a major challenge for the scientific community, given the continuous degradation caused by natural, anthropogenic and climatic factors. The generation of high-resolution 3D documentation is important in the diagnosis of deterioration in historic buildings and the planning of conservation and restoration efforts. The present study proposes an integrated, multi-source workflow combining terrestrial laser scanning (TLS), unmanned aerial vehicle (UAV) photogrammetry, and 3D camera interior scanning. This workflow was employed to document and evaluate the Casa Rusănescu monument in Craiova, Romania. The following processes were incorporated: coordinated acquisition, processing, alignment, evaluation of geometric consistency and deviation-based diagnosis. The diagnosis process include measuring the distance between data clouds and analyzing surface roughness, curvature, planarity and linearity. The workflow was designed to be applicable in real urban conditions, ensuring the coverage of façades, interiors and roof structures. The final, combined dataset contained over 235 million points and includes both interior and exterior geometries. This process helped identify various types of damage, such as cracks, exfoliation, plaster detachment, moisture-related changes, and geometric deformations. An additional AI-assisted validation step (Twinspect) was used to cross-check the degradation indicators derived from point-cloud analyses. The findings suggest that using multiple sensors improves spatial completeness, enhances anomaly detection, and establishes a reliable baseline prior to restoration interventions and long-term monitoring. This methodology facilitates the development of digital twins and GIS-based risk assessments, thereby providing a scalable solution for heritage preservation. Full article

LiDAR sensors are becoming more common and are going to be widely adopted in vehicles in the future by reducing the production cost of the time-of-flight units. Manufacturers are uncertain about the placement, cover material, and shape of the assembly to achieve the optimal performance of the LiDAR, especially in rainy conditions. Although there are existing methodologies for evaluating the visibility and signal intensity of point clouds, there are no indexing approaches available since they would require a broad and comprehensive dataset and realistic and repeatable conditions to perform parametric studies. A matrix of rain conditions with quantified raindrop distribution characteristics is simulated using a wind tunnel via the wind-driven rain concept to produce the realistic impact of raindrops onto the sensor assembly surface at various wind speeds. This paper presents a performance prediction model method for LiDAR sensors and showcases the capability of such a model to provide insights quantitatively when comparing variations. The model is 3-dimensional, including rain conditions perceived by a moving vehicle at different speeds, material properties of surface wettability, and LiDAR visibility in rain compared to dry conditions. The observed LiDAR signal degradation follows an exponential manner, for which this study provides experimentally derived coefficients, enabling quantitative prediction across materials, topologies, rain, and driving speed conditions. Full article

The growing diversity and dynamics of urban environments demand 3D semantic segmentation methods that can recognize a wide range of objects without relying on predefined classes or time-consuming labelled training data. As urban scenes evolve and application requirements vary across locations, flexible, annotation-free 3D segmentation methods are becoming increasingly desirable for large-scale 3D analytics. This work presents the first training-free, open-vocabulary (OV) method for 3D aerial point cloud classification and benchmarks it against state-of-the-art supervised 3D neural networks for the semantic enrichment of these geospatial data. The proposed approach leverages open-vocabulary object recognition in multiple 2D imagery and subsequently projects and refines these detections in 3D space, enabling semantic labelling without prior class definitions or annotated data. In contrast, the supervised baselines are trained on labelled datasets and restricted to a fixed set of object categories. We evaluate all methods with quantitative metrics and qualitative analysis, highlighting their respective strengths, limitations and suitability for scalable urban 3D mapping. By removing the dependency on annotated data and fixed taxonomies, this work represents a key step toward adaptive, scalable and semantic understanding of 3D urban environments. Full article

The intricate layout of underground mine tunnels and the operation of large-scale mining equipment create extensive blind zones, leading to an average of 17 personnel collision accidents per 100,000 working hours in China’s metal mines. To tackle this issue, we constructed a specialized Jinchuan Underground Mining Personnel Dataset covering intersecting tunnels and long straight tunnels, with precise bounding box annotations for personnel locations under varying illumination and dust conditions. We propose the Attention-Augmented PointPillars for Enhanced Mining Personnel Detection. Incorporating Recursive Gated Convolutions into the feature extraction network enables long-range modeling and higher-order spatial interactions. Moreover, the pyramidal design in gⁿ Conv with channel width gradually increasing during spatial interactions enhances the model’s efficiency in processing complex spatial information. Additionally, a Channel and Spatial Attention module integrating spatial and channel attention feature fusion strengthens feature expression via multiple weighting mechanisms. Field tests in Jinchuan underground mine show optimal performance with a batch size of 8, a learning rate of 0.003, and a spatial interaction order of 5, achieving 3% higher accuracy than the original network. Furthermore, comparisons with mainstream methods on the Underground Personnel Dataset confirm our method’s state-of-the-art performance. Full article

Loop closure detection remains a critical challenge in LiDAR-based SLAM, particularly for achieving robust place recognition in environments with rotational and translational variations. To extract more concise environmental representations from point clouds and improve extraction efficiency, this paper proposes a novel composite descriptor—the vertical feature and circle combined (VCC) descriptor, a novel 3D local descriptor designed for efficient and rotation-invariant place recognition. The VCC descriptor captures environmental structure by extracting vertical features from voxelized point clouds and encoding them into circular arc-based histograms, ensuring robustness to viewpoint changes. Under the same hardware, experiments conducted on different datasets demonstrate that the proposed algorithm significantly improves both feature representation efficiency and loop closure recognition performance when compared with the other descriptors, completing loop closure retrieval within 30 ms, which satisfies real-time operation requirements. The results confirm that VCC provides a compact, efficient, and rotation-invariant representation suitable for LiDAR-based SLAM systems. Full article

In cloud computing, ciphertext-policy attribute-based encryption (CP-ABE) is widely adopted for secure data storage and flexible fine-grained access control. For collaborative scenarios involving hierarchical file structures, file hierarchy CP-ABE (FH-CPABE) schemes have been proposed. However, existing file hierarchy CP-ABE schemes rely on computationally intensive bilinear pairing operations, resulting in high overhead. To address this issue, this paper proposes ECC-FH-CPABE, a lightweight and efficient file hierarchy CP-ABE scheme based on elliptic curve cryptography (ECC). By replacing bilinear pairings with scalar multiplication on elliptic curve points, our scheme achieves superior computational efficiency while reducing communication overhead. To ensure strong security while maintaining lightweight performance, this scheme introduces ECC-based data noise to resist user collusion attacks. In addition, ECC-FH-CPABE supports cross-domain data sharing with efficient batch operations, relieving performance bottlenecks. Security analysis proves that the scheme is secure against chosen-plaintext attacks. Extensive simulation results show that ECC-FH-CPABE significantly improves both computational efficiency and communication efficiency compared to existing schemes. Full article

To address fragile feature representation in sparse regions and detail loss in occluded scenes caused by uneven sampling density in 3D point cloud semantic segmentation on the SemanticKITTI dataset, this article proposes an innovative framework that integrates density-adaptive feature enhancement with lightweight spectral fine-tuning, which involves frequency-domain transformations (e.g., Fast Fourier Transform) applied to point cloud features to optimize computational efficiency and enhance robustness in sparse regions, which involves frequency-domain transformations to optimize features efficiently. The method begins by accurately calculating each point’s local neighborhood density using KD tree radius search, subsequently injecting this as an additional feature channel to enable the network’s adaptation to density variations. A density-aware loss function is then employed, dynamically adjusting the classification loss weights—by approximately 40% in low-density areas—to strongly penalize misclassifications and enhance feature robustness from sparse points. Additionally, a multi-view projection fusion mechanism is introduced that projects point clouds onto multiple 2D views, capturing detailed information via mature 2D models, with the primary focus on semantic segmentation tasks using the SemanticKITTI dataset to ensure task specificity. This information is then fused with the original 3D features through backprojection, thereby complementing geometric relationships and texture details to effectively alleviate occlusion artifacts. Experiments on the SemanticKITTI dataset for semantic segmentation show significant performance improvements over the baseline, achieving Precision 0.91, Recall 0.89, and F1-Score 0.90. In low-density regions, the F1-Score improved from 0.73 to 0.80. Ablation studies highlight the contributions of density feature injection, multi-view fusion, and density-aware loss, enhancing F1-Score by 3.8%, 2.5%, and 5.0%, respectively. This framework offers an effective approach for accurate and robust point cloud analysis through optimized density techniques and spectral domain fine-tuning. Full article

In this study, we investigated the effects of flight and processing parameters on the accuracy of UAV-based photogrammetric digital terrain models (DTM) generated from RGB imagery in ageing deciduous and mixed forest stands. Four 100 × 100 m sample plots were selected, for which the reference terrain surface was established using terrestrial laser scanning. Photogrammetric DTMs derived from various parameter combinations were compared against this reference, analysing the magnitude of deviations and the influence of individual parameters through SHAP (SHapley Additive exPlanations) analysis. Based on the identified effects, we provide recommendations for optimal workflows and parameter settings. The processing chain also incorporates a targeted raster-level smoothing procedure developed by the authors, which effectively removes DTM errors caused by point cloud noise left by filtering algorithms, thereby reducing extreme deviations from the reference surface. The results show that the absolute mean elevation error is primarily influenced by flight parameters and ground point classification scale (parameter of the lasground algorithm). Optimal flight parameters were determined at a flight altitude of 100 m, with 80% front and 90% side overlap. Furthermore, a ground classification scale of 9 m proved optimal in forested environments. The proposed targeted smoothing significantly reduced extreme errors, yielding DTMs with a mean error of approximately 6 cm and maximum deviations of about 40 cm. These accuracies demonstrate that UAV-based photogrammetry, when carefully parameterised, provides a reliable basis for surface model normalization and subsequent forest structural analyses. Full article

Ships target detection plays a crucial role in safeguarding maritime transportation. However, affected by factors such as ocean waves, extreme weather, and target diversity (e.g., large size differences, arbitrary rotation, and occlusion), existing deep learning-based detection methods struggle to achieve a satisfactory balance among accuracy, speed, and model size in complex marine environments. To address this challenge, this paper proposes a real-time ship detection algorithm (C-YOLO) integrating global perception and multi-scale feature enhancement. First, a Transformer encoder is added before the detection head, which suppresses interference from sea clutter and cloud mist occlusion through long-range dependency modeling, improving the detection of small and occluded ships. Second, a Dual-Effect Focused Residual Fusion Module is designed to replace the backbone’s multi-scale pooling structure, combining the advantages of CBAM (background noise suppression) and SK-Net (dynamic scale adaptation) to simultaneously capture features of ships of different sizes. Finally, a CZIoU loss function is proposed, which integrates constraints on angle, center point, vertex, and area to address rotation, deformation, and multi-scale issues in ship detection. Experimental results on the SeaShips 7000 dataset show that the proposed C-YOLO achieves a Recall of 0.842, mAP@50 of 0.797, and mAP@50:95 of 0.552, outperforming mainstream algorithms such as YOLOv7 (Recall = 0.785, mAP@50 = 0.781), YOLOv9s (Recall = 0.819, mAP@50 = 0.755), and SSD (Recall = 0.802, mAP@50 = 0.833). With 76.75 M parameters and an inference speed of 119 FPS, the model maintains efficient real-time performance while ensuring detection accuracy. This method effectively reduces false detection and missed detection rates in complex scenarios such as port monitoring and maritime traffic control, providing a reliable technical solution for intelligent maritime surveillance and safe navigation—with significant practical value for improving maritime transportation efficiency and reducing safety risks. Full article