From Virtuality To Reality: A Learning-based Point Cloud Labeling Method With Synthesis Scene

Using mobile Light Detection and Ranging point clouds to accomplish road scene labeling tasks shows promise for a variety of applications. Most existing methods for semantic labeling of point clouds require a huge number of fully supervised point cloud scenes, where each point needs to be manually annotated with a specific category. Manually annotating each point in point cloud scenes is labor intensive and hinders practical usage of those methods. To alleviate such a huge burden of manual annotation, in this paper, we introduce an active learning method that avoids annotating the whole point cloud scenes by iteratively annotating a small portion of unlabeled supervoxels and creating a minimal manually annotated training set. In order to avoid the biased sampling existing in traditional active learning methods, a neighbor-consistency prior is exploited to select the potentially misclassified samples into the training set to improve the accuracy of the statistical model. Furthermore, lots of methods only consider short-range contextual information to conduct semantic labeling tasks, but ignore the long-range contexts among local variables. In this paper, we use a higher order Markov random field model to take into account more contexts for refining the labeling results, despite of lacking fully supervised scenes. Evaluations on three data sets show that our proposed framework achieves a high accuracy in labeling point clouds although only a small portion of labels is provided. Moreover, comparative experiments demonstrate that our proposed framework is superior to traditional sampling methods and exhibits comparable performance to those fully supervised models.

Electroencephalographic (EEG) signal-based personal identification systems have notable advantages and disadvantages. These systems heavily rely on the stability of EEG signals, which several factors. One of the primary factors affecting the stability of EEG signals is an individual's emotional state. Among emotional states, stress significantly impairs people's ability to perform daily tasks. This research aims to identify stress levels, classified as low (2) and high (1), using features from Independent Component Analysis (ICA) and Principal Component Analysis (PCA). Machine learning methods, including Decision Tree, k-Nearest Neighbors (k-NN), Naive Bayes, Support Vector Machine (SVM), and Ensemble techniques, are employed to classify the stress levels. The dataset comprises 40 EEG recordings from the Stroop color-word test, and the data is split using a random holdout function with a ratio of 80% for training and 20% for testing. This study examines the most effective features for identifying stress levels and compares the performance of various machine learning models. The experimental results demonstrate that PCA is the most effective feature extraction method, achieving an average accuracy of 0.718 in stress level classification. Among the machine learning models tested, the Ensemble method performs the best, achieving an accuracy of 0.770 when using PCA features and 0.745 with ICA features. This study highlights the importance of selecting optimal features and machine learning techniques for improving stress detection in EEG-based systems. Further improvements in classification accuracy may be achieved by incorporating additional physiological signals or refining feature extraction techniques.

Multi-spectral Light Detection and Ranging (LiDAR) is an active sensor that can simultaneously obtain spatial geometric data and multi-wavelength information, offering multi-attribute features for geoinformation products such as point cloud scene recognition. Despite these advantages, semantic segmentation of fine-grained point cloud scenes in multi-spectral LiDAR data remains a challenging task, mainly due to the disordered arrangement of point cloud and the large amount of data involved. In particular, it is crucial to extract high-quality features from unstructured point cloud for semantic segmentation. To address these challenges, this paper proposes a Multi-scale Kernel Aggregation Network (MKANet) for point cloud semantic segmentation and evaluates its performance in real scenes. Specifically, MKANet is similar to convolutional kernel operation, which extracts point cloud features by establishing correlation coefficients between point cloud and kernel point, and then aggregates point cloud features into kernels according to coefficient weights. The designed MKANet is a polyhedral structure composed of multiple kernel points, and its interior is multi-layer distributed kernel points. Moreover, the kernel points aggregate the features by the distance weight from the surrounding point cloud. Therefore, multi-scale kernel points provide a more diverse distribution and also provide the feasibility of aggregating features more widely. In addition, multi-scale kernel aggregation can provide more kernel points with different distributions to reduce the confusion caused by the weight contribution of similar points. To further objectively evaluate the effectiveness and performance of MKANet, we conduct a series of comparative experiments using real-world multispectral LiDAR point cloud data. The results of these experiments highlight the critical and influential role of MKANet in scene recognition of multispectral LiDAR point cloud and have reached numerical values of performance.

UA_L-DoTT (University of Alabama’s Large Dataset of Trains and Trucks) is a collection of camera images and 3D LiDAR point cloud scans from five different data sites. Four of the data sites targeted trains on railways and the last targeted trucks on a four-lane highway. Low light conditions were present at one of the data sites showcasing unique differences between individual sensor data. The final data site utilized a mobile platform which created a large variety of viewpoints in images and point clouds. The dataset consists of 93,397 raw images, 11,415 corresponding labeled text files, 354,334 raw point clouds, 77,860 corresponding labeled point clouds, and 33 timestamp files. These timestamps correlate images to point cloud scans via POSIX time. The data was collected with a sensor suite consisting of five different LiDAR sensors and a camera. This provides various viewpoints and features of the same targets due to the variance in operational characteristics of the sensors. The inclusion of both raw and labeled data allows users to get started immediately with the labeled subset, or label additional raw data as needed. This large dataset is beneficial to any researcher interested in machine learning using cameras, LiDARs, or both. The current dataset is publicly available at UA_L-DoTT.

Diabetes mellitus is a metabolic disease that is ranked among the top 10 causes of death by the world health organization. During the last few years, an alarming increase is observed worldwide with a 70% rise in the disease since 2000 and an 80% rise in male deaths. If untreated, it results in complications of many vital organs of the human body which may lead to fatality. Early detection of diabetes is a task of significant importance to start timely treatment. This study introduces a methodology for the classification of diabetic and normal people using an ensemble machine learning model and feature fusion of Chi-square and principal component analysis. An ensemble model, logistic tree classifier (LTC), is proposed which incorporates logistic regression and extra tree classifier through a soft voting mechanism. Experiments are also performed using several well-known machine learning algorithms to analyze their performance including logistic regression, extra tree classifier, AdaBoost, Gaussian naive Bayes, decision tree, random forest, and k nearest neighbor. In addition, several experiments are carried out using principal component analysis (PCA) and Chi-square (Chi-2) features to analyze the influence of feature selection on the performance of machine learning classifiers. Results indicate that Chi-2 features show high performance than both PCA features and original features. However, the highest accuracy is obtained when the proposed ensemble model LTC is used with the proposed feature fusion framework-work which achieves a 0.85 accuracy score which is the highest of the available approaches for diabetes prediction. In addition, the statistical T-test proves the statistical significance of the proposed approach over other approaches.

Detection of banana plants and their major diseases through aerial images and machine learning methods: A case study in DR Congo and Republic of Benin

Accurate and effective classification of lidar point clouds with discriminative features expression is a challenging task for scene understanding. In order to improve the accuracy and the robustness of point cloud classification based on single point features, we propose a novel point set multi-level aggregation features extraction and fusion method based on multi-scale max pooling and latent Dirichlet allocation (LDA). To this end, in the hierarchical point set feature extraction, point sets of different levels and sizes are first adaptively generated through multi-level clustering. Then, more effective sparse representation is implemented by locality-constrained linear coding (LLC) based on single point features, which contributes to the extraction of discriminative individual point set features. Next, the local point set features are extracted by combining the max pooling method and the multi-scale pyramid structure constructed by the point’s coordinates within each point set. The global and the local features of the point sets are effectively expressed by the fusion of multi-scale max pooling features and global features constructed by the point set LLC-LDA model. The point clouds are classified by using the point set multi-level aggregation features. Our experiments on two scenes of airborne laser scanning (ALS) point clouds—a mobile laser scanning (MLS) scene point cloud and a terrestrial laser scanning (TLS) scene point cloud—demonstrate the effectiveness of the proposed point set multi-level aggregation features for point cloud classification, and the proposed method outperforms other related and compared algorithms.

Data collection has become one of the most important aspects of daily life. In visualization of data, the number of points determines the clarity of data one can show. One such resource is a point cloud which is a data structure that represents a collection of huge amounts of multidimensional points and is used to store three-dimensional data. LiDAR is an active optical sensing device that provides extremely accurate x, y and z coordinates of the object it is reflected from. It transmits laser beams to a specific object and reflects its movements back to the receiver and analyzes the time span and measures the distance. The information obtained is used to construct a 3D point cloud of reflected obstacles. Also, by adding color information to the point cloud, we can be converted it from 3D to 4D. They are very dense and may be difficult to analyze since the point clouds can store massive amount of data. Handling such amount of data can be difficult and consumes a considerable amount of time for its visualization. We have designed a 3D Annotation Tool which is an application designed to annotate objects in a point cloud scene. We have annotated the objects by drawing 3D bounding boxes around them. The tool we have designed is capable of visualizing point cloud and RGB images together. It makes accessing and understanding of data easy for the observer. We have used PyQt package to design the GUI interface of the tool. PyQt is a Python binding of the cross-platform GUI toolkit Qt, implemented as a Python plugin. It provides features such as loading of point cloud and RGB images, rotation of RGB image, movement in point cloud and many more features.

A comparative study of PCA, SIMCA and Cole model for classification of bioimpedance spectroscopy measurements

3D point-cloud upsampling, a crucial perceptual module to help us understand the complex scene and object, aims to generate a high-resolution point cloud given a sparse point set. While considerable attention has been paid to single object point-cloud upsampling, literature on upsampling complex scenes has emerged slowly. Remarkably, few related works target LiDAR-based point clouds, which inherently have a density imbalance problem over distance. This paper proposed LiUpNet, a LiDAR-based point cloud upsampling network. Given a sparse and imbalanced point cloud of a natural and complex scene, LiUpNet extracts robust regional features and generates a uniformly distributed dense point cloud that preserves the fine-grained structural architecture. Specifically, in LiUpNet, an attentive and transformer-based feature extractor is applied to learn the detailed regional representation to model the underlying complex local structure. Also, a novel densityinvariant feature consistency loss is introduced to improve the robustness of the learned features against the sparsity changes. Finally, given the regional representations, we append a manifold-based upsampler to super-resolve the regional point clouds up to different scales according to the sparsity and implicitly regresses non-linear 3D scene surfaces. A more uniformly distributed and denser scene point cloud is then achieved with the scheme of dynamic regional upsampling rates. The experimental results on single-object and scene point-cloud upsampling show that LiUpNet outperforms related works qualitatively and quantitatively. Additionally, we demonstrate that the enhanced scene point clouds can efficiently improve the performance of downstream tasks such as point-cloud-based 3D object detection and depth completion.

SnapshotNet: Self-supervised feature learning for point cloud data segmentation using minimal labeled data

Manually annotating complex scene point cloud datasets is both costly and error-prone. To reduce the reliance on labeled data, we propose a snapshot-based self-supervised method to enable direct feature learning on the unlabeled point cloud of a complex 3D scene. A snapshot is defined as a collection of points sampled from the point cloud scene. It could be a real view of a local 3D scan directly captured from the real scene, or a virtual view of such from a large 3D point cloud dataset. First the snapshots go through a self-supervised pipeline including both part contrasting and snapshot clustering for feature learning. Then a weakly-supervised approach is implemented by training a standard SVM classifier on the learned features with a small fraction of labeled data. We evaluate the weakly-supervised approach for point cloud classification by using varying numbers of labeled data and study the minimal numbers of labeled data for a successful classification. Experiments are conducted on three public point cloud datasets, and the results have shown that our method is capable of learning effective features from the complex scene data without any labels.

In point-cloud scenes, semantic segmentation is the basis for achieving an understanding of a 3D scene. The disorderly and irregular nature of 3D point clouds makes it impossible for traditional convolutional neural networks to be applied directly, and most deep learning point-cloud models often suffer from an inadequate utilization of spatial information and of other related point-cloud features. Therefore, to facilitate the capture of spatial point neighborhood information and obtain better performance in point-cloud analysis for point-cloud semantic segmentation, a multiscale, multi-feature PointNet (MSMF-PointNet) deep learning point-cloud model is proposed in this paper. MSMF-PointNet is based on the classical point-cloud model PointNet, and two small feature-extraction networks called Mini-PointNets are added to operate in parallel with the modified PointNet; these additional networks extract multiscale, multi-neighborhood features for classification. In this paper, we use the spherical neighborhood method to obtain the local neighborhood features of the point cloud, and then we adjust the radius of the spherical neighborhood to obtain the multiscale point-cloud features. The obtained multiscale neighborhood feature point set is used as the input of the network. In this paper, a cross-sectional comparison analysis is conducted on the Vaihingen urban test dataset from the single-scale and single-feature perspectives.

Three-dimensional laser point cloud technology has been widely used in machine vision and three-dimensional reconstruction engineering, but scene recognition and semantic segmentation of three-dimensional point cloud is still developing and improving. With the gradual maturity of artificial neural networks of deep learning in the field of computer vision, it is possible to identify and segment three-dimensional point cloud scenes. However, most methods perform slowly and inaccurately in real-world outdoor scenes that are large and noisy. This paper proposes a new scenario segmentation method based on PointNet and VoxelNet to improve the efficiency and accuracy of semantic segmentation. Firstly, the voxel meshing method is used to downsample the point cloud scene while keeping the original information of the point cloud. Then the point cloud data is filtered by multiple methods to remove noises and smooth the surface of point clouds. Finally, the processed point cloud scene is trained and the ideal semantic segmentation result is obtained. The results of our method are evaluated on the datasets of BDCI and our own datasets.

The 3D reconstruction technology based on point cloud can accurately present various elements and facilities in the environment in the digital city model, and realize the visual management and operation of smart city.In this paper, a monochromatic rhomboid calibration object is proposed to reconstruct the 3D point cloud scene. Firstly, the comprehensive environmental point cloud information is obtained by rotating scanning. Then the point cloud of the calibration object is extracted based on the threshold segmentation of coordinate axes and Random Sample Consensus (RANSAC) algorithm. In order to accurately extract the feature points of the calibration object, the point clouds scanned by different laser beam were clustered, and then the boundary points were calculated to generate the scene feature points. Finally, Iterative Closest Point (ICP) algorithm was used to register feature points of different scenes to obtain the inter-frame transformation matrix, so as to complete the high-precision point cloud scene reconstruction and realize the high-precision digital presentation of the real world scene.