Sequential Point Research Articles

Unsupervised 3D Object Segmentation of Point Clouds by Geometry Consistency.

In this paper, we study the problem of 3D object segmentation from raw point clouds. Unlike existing methods which usually require a large amount of human annotations for full supervision, we propose the first unsupervised method, called OGC, to simultaneously identify multiple 3D objects in a single forward pass, without needing any type of human annotations. The key to our approach is to fully leverage the dynamic motion patterns over sequential point clouds as supervision signals to automatically discover rigid objects. Our method consists of three major components, 1) the object segmentation network to directly estimate multi-object masks from a single point cloud frame, 2) the auxiliary self-supervised scene flow estimator, and 3) our core object geometry consistency component. By carefully designing a series of loss functions, we effectively take into account the multi-object rigid consistency and the object shape invariance in both temporal and spatial scales. This allows our method to truly discover the object geometry even in the absence of annotations. We extensively evaluate our method on five datasets, demonstrating the superior performance for object part instance segmentation and general object segmentation in both indoor and the challenging outdoor scenarios.

Bioinformatics design of peptide binding to the human cardiac troponin I (cTnI) in biosensor development for myocardial infarction diagnosis.

Cardiovascular disease has reached a mortality rate of 470,000 patients each year. Myocardial infarction accounts for 49.2% of these deaths, and the cTnI protein is a crucial target in diagnosing myocardial infarction. A peptide-based bioreceptor design using a computational approach is a good candidate to be developed for a rapid, effective, and selective detection method for cTnI although it is still lacking in study. Hence, to address the scientific gap, we develop a new candidate peptide for the cTnI biosensor by bioinformatics method and present new computational approaches. The sequential point mutations were made to the selected peptide to increase its stability and affinity for cTnI. Next, molecular docking was performed to select the mutated peptide, and one of the best results was subjected to the molecular dynamics simulation. Finally, the results showed that the best peptide showed the lowest affinity and good stability among other mutated peptide designs for interacting with the cTnI protein. In addition, the peptide has been tested to have a higher specificity towards cTnI than its major isomer, sTnI, through molecular docking and molecular dynamics simulation. Therefore, the peptide is considered a good potential bioreceptor for diagnosing myocardial infarction diseases.

Two-stage and purely sequential minimum risk point estimation of the scale parameter of a family of distributions under modified LINEX loss plus sampling cost

Two-stage and purely sequential minimum risk point estimation of the scale parameter of a family of distributions under modified LINEX loss plus sampling cost

LiDAR‐based place recognition for mobile robots in ground/water surface multiple scenes

AbstractLiDAR‐based 3D place recognition is an essential component of simultaneous localization and mapping systems in multi‐scene robotic applications. However, extracting discriminative and generalizable global descriptors of point clouds is still an open issue due to the insufficient use of the information contained in the LiDAR scans in existing approaches. In this paper, we propose a novel spatial‐temporal point cloud encoding network for multiple scenes, dubbed STM‐Net, to fully fuse the multi‐view spatial information and temporal information of LiDAR point clouds. Specifically, we first develop a spatial feature encoding module consisting of the single‐view transformer and multi‐view transformer. The module learns the correlation both within a single view and between two views by utilizing the multi‐layer range images generated by spherical projection and multi‐layer bird's eye view images generated by top‐down projection. Then in the temporal feature encoding module, we exploit the temporal transformer to mine the temporal information in the sequential point clouds, and a NetVLAD layer is applied to aggregate features and generate sub‐descriptors. Furthermore, we use a GeM pooling layer to fuse more information along the time dimension for the final global descriptors. Extensive experiments conducted on unmanned ground/surface vehicles with different LiDAR configurations indicate that our method (1) achieves superior place recognition performance than state‐of‐the‐art algorithms, (2) generalizes well to diverse sceneries, (3) is robust to viewpoint changes, (4) can operate in real‐time, demonstrating the effectiveness and satisfactory capability of the proposed approach and highlighting its promising applications in multi‐scene place recognition tasks.

Sequential Point Clouds: A Survey.

Point clouds have garnered increasing research attention and found numerous practical applications. However, many of these applications, such as autonomous driving and robotic manipulation, rely on sequential point clouds, essentially adding a temporal dimension to the data (i.e., four dimensions) because the information of the static point cloud data could provide is still limited. Recent research efforts have been directed towards enhancing the understanding and utilization of sequential point clouds. This paper offers a comprehensive review of deep learning methods applied to sequential point cloud research, encompassing dynamic flow estimation, object detection & tracking, point cloud segmentation, and point cloud forecasting. This paper further summarizes and compares the quantitative results of the reviewed methods over the public benchmark datasets. Ultimately, the paper concludes by addressing the challenges in current sequential point cloud research and pointing towards promising avenues for future research.

Sequential good lattice point sets for computer experiments

Sequential Latin hypercube designs have recently received great attention for computer experiments. Much of the work has been restricted to invariant spaces. The related systematic construction methods are inflexible while algorithmic methods are ineffective for large designs. For such designs in space contraction, systematic construction methods have not been investigated yet. This paper proposes a new method for constructing sequential Latin hypercube designs via good lattice point sets in a variety of experimental spaces. These designs are called sequential good lattice point sets. Moreover, we provide fast and efficient approaches for identifying the (nearly) optimal sequential good lattice point sets under a given criterion. Combining with the linear level permutation technique, we obtain a class of asymptotically optimal sequential Latin hypercube designs in invariant spaces where the $L_1$-distance in each stage is either optimal or asymptotically optimal. Numerical results demonstrate that the sequential good lattice point set has a better space-filling property than the existing sequential Latin hypercube designs in the invariant space. It is also shown that the sequential good lattice point sets have less computational complexity and more adaptability.

Phase 2 study of xevinapant + chemoradiotherapy (CRT) vs placebo + CRT in patients with unresected locally advanced squamous cell carcinoma of the head and neck (LA SCCHN): A post hoc activity analysis by blinded independent review committee evaluation.

e18039 Background: The randomized, placebo-controlled, double-blind, phase 2 study (NCT02022098) of xevinapant + CRT demonstrated for the first time improved efficacy outcomes, including 5-year overall survival, when adding a new treatment modality to CRT for patients (pts) with unresected LA SCCHN. Here, we report a post hoc analysis comparing selected efficacy endpoints by blinded independent review committee (BIRC) evaluation with previously reported outcomes by investigator analysis. Methods: The study design was previously published (Sun X-S, et al. Lancet Oncol. 2020;21[9]:1173-87). Pts received oral xevinapant 200 mg/day (Days 1–14 of a 21-day cycle for 3 cycles), or matched placebo, with standard-of-care cisplatin-based CRT. Progression-free survival (PFS), duration of response (DOR), best overall response (BOR), and time to response (TTR) were independently evaluated by a two-reader, sequential time point presentation, batch read mode paradigm. Radiologists were blinded to investigator assessments. Discrepancies were adjudicated by a blinded, independent third radiologist. Results: All 96 pts randomized into the study (n=48 in each arm) were evaluated by BIRC. Response concordance between BIRC and investigator assessment was 84.4%. There were 31 PFS events by BIRC (xevinapant + CRT, 11 events; placebo + CRT, 20 events) and 38 PFS events by investigator (xevinapant + CRT, 12 events; placebo + CRT, 26 events) assessment, respectively. Xevinapant + CRT reduced the risk of disease progression (by BIRC) or death vs placebo + CRT by 62% (HR, 0.38 [95% CI, 0.18–0.79]; p=0.0099); median PFS was not reached (NR) vs 20.8 months, respectively. PFS outcomes by investigator analysis (HR, 0.33; 95% CI, 0.17–0.67; p=0.0019) were similar. DOR (by BIRC) was prolonged with xevinapant + CRT vs placebo + CRT (median, NR vs 31.4 months; HR, 0.31 [95% CI, 0.11–0.83]; p=0.0196), which was similar to outcomes by investigator analysis (adjusted HR, 0.21; 95% CI, 0.08–0.54; p=0.0011). Complete response (CR; by BIRC) was the most frequent BOR in both arms (xevinapant + CRT, 28 [58.3%]; placebo + CRT, 19 [39.6%]); the CR rate was greater in pts treated with xevinapant + CRT vs pts treated with placebo + CRT (odds ratio, 2.56; 95% CI, 1.07–6.13; p=0.0344). The CR rate trends by BIRC evaluation were similar to those by investigator analysis (xevinapant + CRT, 31 [64.4%]; placebo + CRT, 26 [54.2%]). A difference in TTR (by BIRC) favoring xevinapant + CRT over placebo + CRT was also observed (HR, 0.65 [95% CI, 0.39–1.07]). Conclusions: While fewer PFS events were observed by BIRC than by investigator assessment, the BIRC results were generally consistent with those of the investigators, with only minor differences in endpoint outcomes. Clinical trial information: NCT02022098 .

Online Detection and Fuzzy Clustering of Anomalies in Non-Stationary Time Series

We consider the challenge of detecting and clustering point and collective anomalies in streaming data that exhibit significant nonlinearities and seasonal structures. The challenge is motivated by detecting problems in a communications network, where we can measure the throughput of nodes, and wish to rapidly detect anomalous traffic behaviour. Our approach is to train a neural network-based nonlinear autoregressive exogenous model on initial training data, then to use the sequential collective and point anomaly framework to identify anomalies in the residuals generated by comparing one-step-ahead predictions of the fitted model with the observations, and finally, we cluster the detected anomalies with fuzzy c-means clustering using empirical cumulative distribution functions. The autoregressive model is sufficiently general and robust such that it provides the nearly (locally) stationary residuals required by the anomaly detection procedure. The combined methods are successfully implemented to create an adaptive, robust, computational framework that can be used to cluster point and collective anomalies in streaming data. We validate the method on both data from the core of the UK’s national communications network and the multivariate Skoltech anomaly benchmark and find that the proposed method succeeds in dealing with different forms of anomalies within the nonlinear signals and outperforms conventional methods for anomaly detection and clustering.

Design and Evaluation of Field Segmentation, Path Generation and Sequential Point Tracking Algorithms for Agricultural Mobile Robots

Highlights Field segmentation, path generation, and sequential point tracking algorithms were designed for agricultural mobile robots. An optimized field segmentation plan and operation path could be generated for the robot. The sequential point tracking algorithm could guide the robot to a series of target points. The algorithms were tested in simulation calculations and practical experiments. Abstract. As agricultural production becomes increasingly intelligent, operations by autonomous mobile robots become an inevitable trend. For some operations, such as in-field soil sampling, agricultural robots need to track a series of target points in the field. In this work, a field segmentation algorithm, a path generation algorithm, and a sequential point tracking algorithm were designed and evaluated. The field segmentation and operation path generation algorithms, which took, respectively, the total lost area of the field and the total turning angle of the robot as optimization objectives, aimed to provide an efficient operation plan for the robot to follow. The sequential point tracking algorithm was designed for the robot to realize automatic tracking of the planned target points following a desired sequence. The algorithms were tested by simulation calculations using the boundary coordinates of two test fields and field experiments based on a small four-wheeled mobile robot. Field results of the straight-path, multi-target point tracking experiment showed that the designed steering angle and velocity control laws could accurately guide the robot from a start point to a series of target points when the robot was running on a concrete surface. The absolute distance errors of the robot with respect to the targets were smaller than 0.03 m. To test the robot’s tracking performance on different surface conditions, two whole-field, continuous point tracking experiments were performed. The general effectiveness of the sequential point tracking algorithm was validated. When the robot was running on hard surface conditions (concrete and grass), the tracking errors were between 0.014 m and 0.074 m. When the robot was running on a soft and bumpy soil surface, the tracking accuracy degraded to submeter level due to the limitation of small-size wheels. The average, RMS, and maximum errors were 0.774 m, 0.775 m, and 0.798 m, respectively. However, the feasibility of the sequential control algorithm was not influenced. The designed algorithms could be potentially utilized in the development of robotic systems for in-field site-specific operations. Keywords: Agricultural mobile robot, Control sequence, Field segmentation, Operation path generation, Point tracking.

The state of cumulative sum sequential changepoint testing 70 years after Page

Abstract Quality control charts aim at raising an alarm as soon as sequentially obtained observations of an underlying random process no longer seem to be within stochastic fluctuations prescribed by an in-control scenario. Such random processes can often be modelled using the concept of stationarity, or even independence as in most classical works. An important out-of-control scenario is the changepoint alternative, for which the distribution of the process changes at an unknown point in time. In his seminal 1954 Biometrika paper, E. S. Page introduced the famous cumulative sum control charts for changepoint monitoring. Innovatively, decision rules based on cumulative sum procedures took the full history of the process into account, whereas previous procedures were based only on a fixed and typically small number of the most recent observations. The extreme case of using only the most recent observation, often referred to as the Shewhart chart, is more akin to serial outlier than changepoint detection. Page’s cumulative sum approach, introduced seven decades ago, is ubiquitous in modern changepoint analysis, and his original paper has led to a multitude of follow-up papers in different research communities. This review is focused on a particular subfield of this research, namely nonparametric sequential, or online, changepoint tests that are constructed to maintain a desired Type-1 error as opposed to the more traditional approach seeking to minimize the average run length of the procedures. Such tests have originated at the intersection of econometrics and statistics. We trace the development of these tests and highlight their properties, mostly using a simple location model for clarity of exposition, but we also review more complex situations such as regression and time series models.

Sequential GNSS-Acoustic seafloor point positioning with modeling of sound speed variation

Sequential GNSS-Acoustic seafloor point positioning with modeling of sound speed variation

A note on online change point detection

We investigate sequential change point estimation and detection in univariate nonparametric settings, where a stream of independent observations from sub-Gaussian distributions with a common variance factor and piecewise-constant but otherwise unknown means are collected. We develop a simple CUSUM-based methodology that provably control the probability of false alarms or the average run length while minimizing, in a minimax sense, the detection delay. We allow for all the model parameters to vary in order to capture a broad range of levels of statistical hardness for the problem at hand. We further show how our methodology is applicable to the case in which multiple change points are to be estimated sequentially.

Estimation of fractal dimension of trees using LiDAR point data with sequential data decimation

Fractal dimensions of trees provide key ecological information regarding tree and forest stand structure. Although LiDAR scanning has the potential to reduce the intensive labor required to estimate fractal dimensions with manual measurements, neither an accuracy assessment comparing with true dimensions nor a guide or method for accurately measuring the dimensions using LiDAR scanners has been presented. This study examined the accuracy of fractal dimensions estimated using terrestrial LiDAR scanning (TLS) and unmanned aerial LiDAR scanning (ULS) through computer simulations and developed an approach to reduce the estimation bias. The true fractal dimensions of digitized trees of red pine (Pinus densiflora), ubame oak (Quercus phillyreoides), and giant timber bamboo (Phyllostachys bambusoides) in 18 site scenarios were calculated by applying the box-counting method to the tree object data. The true fractal dimensions were compared with the estimated dimensions in four TLS and ULS scenarios. Furthermore, a method to reduce the estimation bias was developed by introducing sequential point decimation and modeling the acquisition rate decay, thereby extrapolating the number of voxels before decay. Simple box-counting on highly dense and precise TLS data collected from four positions could retrieve the true fractal dimensions for all 18 site scenarios. However, sparse and less accurate TLS underestimated fractal dimensions in 8 out of 18 cases, and ULS overestimated the dimensions in 29 of 36 cases. The change in the relationship between the number of voxels and mesh size on a log-log plot with sequential decimation could be used to judge whether a set of TLS data can retrieve accurate fractal dimensions by applying simple box-counting to the TLS data. The developed method improved the estimations of fractal dimensions for 30 of the 37 cases where the simple box-counting on the TLS and ULS data provided biased estimations. In contrast, the estimation bias increased in 27 of the 35 cases where the simple box-counting on the TLS and ULS data successfully retrieved the true fractal dimensions.

Ergodicity for three-dimensional stochastic Navier–Stokes equations with Markovian switching

Asymptotic behavior of the three-dimensional stochastic Navier-Stokes equations with Markovian switching for additive noises is studied for an incompressible fluid flow in a bounded domain. The objective of this article is to prove the existence and identification of a stationary measure for such stochastic systems. To achieve this goal, a family of regularized equations is introduced in order to investigate the asymptotic behavior of their solutions. The existence of an ergodic measure for the regularized system is established by the Krylov-Bogolyubov method. The existence and identification of a stationary measure corresponding to the original system are obtained as a sequential limit point of the family of the ergodic measures that correspond to the regularized systems. The main result of this work is that under mild integrability conditions on the initial data and external force, there exists a stationary measure for the solution of the three-dimensional stochastic Navier-Stokes equations with Markovian switching.

Real-time change point detection in linear models using the ranking selection procedure

We propose a novel sequential change point detection method in linear models. Our method uses a given historical data set to determine the prechange model. Significant features are selected using the ranking procedure, which is an innovative approach aimed at revealing the rank of all features in terms of their effects on the model. We establish the asymptotic properties of the test statistic under the null and alternative hypotheses. Simulations are conducted to illustrate the performance of the proposed method. We conclude with a real data application to illustrate the detection procedure.

Multi-target 3D single-molecule super resolution imaging throughout mammalian cells with a single-objective tilted light sheet.

Single-molecule based super-resolution fluorescence microscopy offers an approach to sub-diffraction-limited imaging of biological samples. While these approaches have proved significant to understanding subcellular structures, numerous super-resolution systems suffer from drawbacks such as decreased localization precision due to high background fluorescence, lack of control of the extracellular environment, decreased multi-target accuracy due to chromatic offsets in multicolor imaging, and the inability to image structures in three dimensions (3D). Light sheet fluorescence microscopy, which involves illuminating the sample with a thin sheet of light, offers a gentle illumination scheme that greatly reduces photobleaching, photodamage, and out of focus background fluorescence, thereby improving the localization precision. Here, we present a single-objective tilted light sheet super-resolution setup with five added innovations: (i) a microfluidic chip with reflective sidewalls and transparent top, (ii) sequential DNA Point Accumulation for Imaging in Nanoscale Topography (DNA-PAINT) known as Exchange-PAINT, (iii) deep learning for high density single emitter localization, (iv) double helix point spread functions for 3D imaging, and (iv) galvanometric dithering of the illumination beam to reduce scattering artifacts and provide homogenous sample illumination. By utilizing light sheet illumination in conjunction with Exchange-PAINT and deep learning, this setup allows for the use of high concentrations of imager strands given that background fluorescence is drastically reduced and sparse emitters are no longer necessary for accurate localization. Furthermore, sequential washes of imager strands are enabled through the use of microfluidics, which allows for accurate, multi-target reconstructions that are not subject to chromatic offsets. Thus, we demonstrate a setup that offers improved localization precision, extracellular environment control, multi-target information, and 3D imaging that can easily be implemented for fast, accurate, and precise multi-target super-resolution whole cell imaging.

DeepPCT: Single Object Tracking in Dynamic Point Cloud Sequences

Three-dimensional single object tracking (SOT) in dynamic point cloud sequences is a fundamental problem for numerous intelligent systems. In this article, we propose a deep neural architecture named DeepPCT to achieve accurate and robust tracking in sequential point clouds. Instead of relying on the bounding box classification and regression to localize the object, the proposed framework directly formulates the intersection-over-union (IoU) prediction as its objectives for accurate 3-D localization. A Center IoU (CIoU) is further proposed to mitigate the ambiguity of standard IoU. In addition, we also investigate the impact of an additional re-detection module to the overall tracking performance, since tracking failures frequently occur due to the large variations and error accumulation. Experimental results on the KITTI benchmark demonstrate that the proposed tracker achieves competitive performance, with a 3-D precision rate of 63.7% and a 3-D success rate of 45.0%. Moreover, our tracker achieves compelling performance in the wild with severe illumination variations and occlusions, and even outperforms several 2-D counterpart trackers.

SemanticFlow: Semantic Segmentation of Sequential LiDAR Point Clouds From Sparse Frame Annotations

Sequential point clouds acquired by light detection and ranging (LiDAR) technology provide accurate spatial information for environmental sensing. However, semantic segmentation of point cloud sequences relies on many manual point-wise annotations, which are error-prone and expensive. Existing mainstream weakly supervised methods tackle this by reducing the percentage of labeled points, but they are mostly designed for static indoor scenes and are hard to apply practically. From the viewpoint of realistic annotation procedures and the nature of point cloud sequences, this paper proposes a novel semantic segmentation method, SemanticFlow, for LiDAR point cloud sequences using sparse frames with annotations. The proposed method achieves competitive performance compared with fully supervised methods. Specifically, we designed a bidirectional cross-frame pseudo label propagation module that uses scene flow to learn the correlation and propagate pseudo labels across neighboring frames. In addition, a label refinement mechanism is proposed to select reliable pseudo labels for learning. Extensive experiments on SemanticKITTI, SemanticPOSS, and Synthia 4D datasets demonstrate that our sparse frame annotation method is compatible with some fully supervised counterparts.

Distribution-free minimum risk point estimation of the mean under powered absolute error loss plus cost of sampling: Illustrations with cancer data

We begin with a discussion of results on the asymptotic properties associated with purely sequential minimum risk point estimation (MRPE) methodology for an unknown mean when the distribution is unspecified, under squared error loss (SEL) plus the cost of sampling. We proceed to show that, given certain moment conditions, such properties can be extended to estimation problems under a broader class of loss functions, namely: first order risk-efficiency under powered absolute error loss (PAEL) plus the powered cost of sampling. In addition to purely sequential strategy, we also prove these asymptotic properties for a more operationally convenient, accelerated sequential procedure. We follow-up with extensive data analyses from simulations under mixture distributions. We conclude with illustrations using data from two cancer studies.

Estimation of forest stand characteristics using individual tree detection, stochastic geometry and a sequential spatial point process model

Airborne Laser Scanning (ALS) results in point-wise measurements of canopy height, which can further be used for Individual Tree Detection (ITD). However, ITD cannot find all trees because small trees can hide below larger tree crowns. Here we discuss methods where the plot totals and means of tree-level characteristics are estimated in such context. The starting point is a previously presented Horvitz–Thompson-like (HT-like) estimator, where the detectability is based on the larger tree crowns and a tuning parameter α that models the detection condition. We propose a new method which is based on modeling the spatial pattern of hidden tree locations using a sequential spatial point process model, with a tuning parameter θ. We also explore whether the variability of the tuning parameters α and θ can be predicted using ALS features to improve the predictions. The accuracy of stand density, dominant height and mean height is used as comparison criteria in a cross-validation procedure. The HT-like estimator with empirically estimated tuning parameter α performed the best. The overall performance of the new method was comparable. The new method was computationally less demanding, which makes it attractive for practical use.