基于Contextual Transformer的自动驾驶单目3D目标检测

针对当前单目3D目标检测中存在的漏检和多尺度目标检测效果不佳的问题,提出了一种基于Contextual Transformer的自动驾驶单目3D目标检测算法(CM-RTM3D)。在ResNet-50网络中引入Contextual Transformer(CoT),构建ResNet-Transformer架构以提取特征。设计多尺度空间感知模块(MSP),通过尺度空间响应操作改善浅层特征的丢失情况,嵌入沿水平和竖直两个空间方向的坐标注意力机制(CA),使用softmax函数生成各尺度的重要性软权重。在偏移损失中采用Huber损失函数代替L1损失函数。实验结果表明:在KITTI自动驾驶数据集上,相较于RTM3D算法,该算法在简单、中等、困难三个难度级别下,AP3D分别提升了4.84、3.82、5.36个百分点,APBEV分别提升了4.75、6.26、3.56个百分点。

Three-dimensional numerical analysis of plant-soil hydraulic interactions on pore water pressure of vegetated slope under different rainfall patterns

Understanding the pore water pressure distribution in unsaturated soil is crucial in predicting shallow landslides triggered by rainfall,mainly when dealing with different temporal patterns of rainfall intensity.However,the hydrological response of vegetated slopes,especially three-dimensional(3D)slopes covered with shrubs,under different rainfall patterns remains unclear and requires further investigation.To address this issue,this study adopts a novel 3D numerical model for simulating hydraulic interactions between the root system of the shrub and the surrounding soil.Three series of numerical parametric studies are conducted to investigate the influences of slope inclination,rainfall pattern and rainfall duration.Four rainfall patterns(advanced,bimodal,delayed,and uniform)and two rainfall durations(4-h intense and 168-h mild rainfall)are considered to study the hydrological response of the slope.The computed results show that 17%higher transpiration-induced suction is found for a steeper slope,which remains even after a short,intense rainfall with a 100-year return period.The extreme rainfalls with advanced(PA),bimodal(PB)and uniform(PU)rainfall patterns need to be considered for the short rainfall duration(4 h),while the delayed(PD)and uniform(PU)rainfall patterns are highly recommended for long rainfall durations(168 h).The presence of plants can improve slope stability markedly under extreme rainfall with a short duration(4 h).For the long duration(168 h),the benefit of the plant in preserving pore-water pressure(PWP)and slope stability may not be sufficient.

Three-dimensional stability calculation method for high and large composite slopes formed by mining stope and inner dump in adjacent open pits

The 2D limit equilibrium method is widely used for slope stability analysis.However,with the advancement of dump engineering,composite slopes often exhibit significant 3D mechanical effects.Consequently,it is of significant importance to develop an effective 3D stability calculation method for composite slopes to enhance the design and stability control of open-pit slope engineering.Using the composite slope formed by the mining stope and inner dump in Baiyinhua No.1 and No.2 open-pit coal mine as a case study,this research investigates the failure mode of composite slopes and establishes spatial shape equations for the sliding mass.By integrating the shear resistance and sliding force of each row of microstrip columns onto the bottom surface of the strip corresponding to the main sliding surface,a novel 2D equivalent physical and mechanical parameters analysis method for the strips on the main sliding surface of 3D sliding masses is proposed.Subsequently,a comprehensive 3D stability calculation method for composite slopes is developed,and the quantitative relationship between the coordinated development distance and its 3D stability coefficients is examined.The analysis reveals that the failure mode of the composite slope is characterized by cutting-bedding sliding,with the arc serving as the side interface and the weak layer as the bottom interface,while the destabilization mechanism primarily involves shear failure.The spatial form equation of the sliding mass comprises an ellipsoid and weak plane equation.The analysis revealed that when the coordinated development distance is 1500 m,the error rate between the 3D stability calculation result and the 2D stability calculation result of the composite slope is less than 8%,thereby verifying the proposed analytical method of equivalent physical and mechanical parameters and the 3D stability calculation method for composite slopes.Furthermore,the3D stability coefficient of the composite slope exhibits an exponential correlation with the coordinated development distance,with the coefficient gradually decreasing as the coordinated development distance increases.These findings provide a theoretical guideline for designing similar slope shape parameters and conducting stability analysis.

Short-term displacement prediction for newly established monitoring slopes based on transfer learning

This study makes a significant progress in addressing the challenges of short-term slope displacement prediction in the Universal Landslide Monitoring Program,an unprecedented disaster mitigation program in China,where lots of newly established monitoring slopes lack sufficient historical deformation data,making it difficult to extract deformation patterns and provide effective predictions which plays a crucial role in the early warning and forecasting of landslide hazards.A slope displacement prediction method based on transfer learning is therefore proposed.Initially,the method transfers the deformation patterns learned from slopes with relatively rich deformation data by a pre-trained model based on a multi-slope integrated dataset to newly established monitoring slopes with limited or even no useful data,thus enabling rapid and efficient predictions for these slopes.Subsequently,as time goes on and monitoring data accumulates,fine-tuning of the pre-trained model for individual slopes can further improve prediction accuracy,enabling continuous optimization of prediction results.A case study indicates that,after being trained on a multi-slope integrated dataset,the TCN-Transformer model can efficiently serve as a pretrained model for displacement prediction at newly established monitoring slopes.The three-day average RMSE is significantly reduced by 34.6%compared to models trained only on individual slope data,and it also successfully predicts the majority of deformation peaks.The fine-tuned model based on accumulated data on the target newly established monitoring slope further reduced the three-day RMSE by 37.2%,demonstrating a considerable predictive accuracy.In conclusion,taking advantage of transfer learning,the proposed slope displacement prediction method effectively utilizes the available data,which enables the rapid deployment and continual refinement of displacement predictions on newly established monitoring slopes.

Efficient single-pixel imaging encrypted transmission based on 3D Arnold transformation

Single-pixel imaging(SPI)can transform 2D or 3D image data into 1D light signals,which offers promising prospects for image compression and transmission.However,during data communication these light signals in public channels will easily draw the attention of eavesdroppers.Here,we introduce an efficient encryption method for SPI data transmission that uses the 3D Arnold transformation to directly disrupt 1D single-pixel light signals and utilizes the elliptic curve encryption algorithm for key transmission.This encryption scheme immediately employs Hadamard patterns to illuminate the scene and then utilizes the 3D Arnold transformation to permutate the 1D light signal of single-pixel detection.Then the transformation parameters serve as the secret key,while the security of key exchange is guaranteed by an elliptic curve-based key exchange mechanism.Compared with existing encryption schemes,both computer simulations and optical experiments have been conducted to demonstrate that the proposed technique not only enhances the security of encryption but also eliminates the need for complicated pattern scrambling rules.Additionally,this approach solves the problem of secure key transmission,thus ensuring the security of information and the quality of the decrypted images.

Depth-Guided Vision Transformer With Normalizing Flows for Monocular 3D Object Detection

Monocular 3D object detection is challenging due to the lack of accurate depth information.Some methods estimate the pixel-wise depth maps from off-the-shelf depth estimators and then use them as an additional input to augment the RGB images.Depth-based methods attempt to convert estimated depth maps to pseudo-LiDAR and then use LiDAR-based object detectors or focus on the perspective of image and depth fusion learning.However,they demonstrate limited performance and efficiency as a result of depth inaccuracy and complex fusion mode with convolutions.Different from these approaches,our proposed depth-guided vision transformer with a normalizing flows(NF-DVT)network uses normalizing flows to build priors in depth maps to achieve more accurate depth information.Then we develop a novel Swin-Transformer-based backbone with a fusion module to process RGB image patches and depth map patches with two separate branches and fuse them using cross-attention to exchange information with each other.Furthermore,with the help of pixel-wise relative depth values in depth maps,we develop new relative position embeddings in the cross-attention mechanism to capture more accurate sequence ordering of input tokens.Our method is the first Swin-Transformer-based backbone architecture for monocular 3D object detection.The experimental results on the KITTI and the challenging Waymo Open datasets show the effectiveness of our proposed method and superior performance over previous counterparts.

Transformation and Breaking of Irregular Waves on Very Gentle Slopes

Based on theoretical analysis, numerical calculation, and experimental study. this paper discusses breaker indices of irregular waves, transformation of wave spectrum, characteristics and computation of breaking waves, as well as the critical beach slope under which waves will not break. Computed results are in good agreement with laboratory physical model test data and ocean wave field measurements.

Seismic dip estimation based on the twodimensional Hilbert transform and its application in random noise attenuation

In seismic data processing, random noise seriously affects the seismic data quality and subsequently the interpretation. This study aims to increase the signal-to-noise ratio by suppressing random noise and improve the accuracy of seismic data interpretation without losing useful information. Hence, we propose a structure-oriented polynomial fitting filter. At the core of structure-oriented filtering is the characterization of the structural trend and the realization of nonstationary filtering. First, we analyze the relation of the frequency response between two-dimensional（2D） derivatives and the 2D Hilbert transform. Then, we derive the noniterative seismic local dip operator using the 2D Hilbert transform to obtain the structural trend. Second, we select polynomial fitting as the nonstationary filtering method and expand the application range of the nonstationary polynomial fitting. Finally, we apply variableamplitude polynomial fitting along the direction of the dip to improve the adaptive structureoriented filtering. Model and field seismic data show that the proposed method suppresses the seismic noise while protecting structural information.

Estimating primaries by sparse inversion of the 3D Curvelet transform and the L1-norm constraint

In this paper, we built upon the estimating primaries by sparse inversion （EPSI） method. We use the 3D curvelet transform and modify the EPSI method to the sparse inversion of the biconvex optimization and Ll-norm regularization, and use alternating optimization to directly estimate the primary reflection coefficients and source wavelet. The 3D curvelet transform is used as a sparseness constraint when inverting the primary reflection coefficients, which results in avoiding the prediction subtraction process in the surface-related multiples elimination （SRME） method. The proposed method not only reduces the damage to the effective waves but also improves the elimination of multiples. It is also a wave equation- based method for elimination of surface multiple reflections, which effectively removes surface multiples under complex submarine conditions.

多尺度Transformer激光雷达点云3D物体检测

激光雷达点云3D物体检测,对于小物体如行人、自行车的检测精度较低,容易漏检误检,提出一种多尺度Transformer激光雷达点云3D物体检测方法 MSPT-RCNN(multi-scale point transformer-RCNN),提高点云3D物体检测精度。该方法包含两个阶段,即第一阶段(RPN)和第二阶段(RCNN)。RPN阶段通过多尺度Transformer网络提取点云特征,该网络包含多尺度邻域嵌入模块和跳跃连接偏移注意力模块,获取多尺度邻域几何信息和不同层次全局语义信息,生成高质量初始3D包围盒;在RCNN阶段,引入包围盒内的点云多尺度邻域几何信息,优化了包围盒位置、尺寸、朝向和置信度等信息。实验结果表明,该方法(MSPT-RCNN)具有较高检测精度,特别是对于远处和较小物体,提升更高。MSPT-RCNN通过有效学习点云数据中的多尺度几何信息,提取不同层次有效的语义信息,能够有效提升3D物体检测精度。

探讨NLR、PLR、hs-CRP、Hb、D-D与ACI非溶栓治疗后出血性转化的相关性分析

目的 探讨急性脑梗死患者(ACI)非溶栓治疗前血清中性粒细胞与淋巴细胞比值(NLR)、血小板与淋巴细胞比值(PLR)、超敏C反应蛋白(hs-CRP)、血红蛋白(Hb)、D-二聚体(D-D)与治疗后出血性转化(HT)的相关性,以期为临床预防ACI非溶栓治疗后发生提供指导依据。方法 本研究病历组(设为转化组)共纳入80例样本,选取自2015年1月至2019年12月在吉林大学第一医院检验科接受非溶栓治疗后并出现HT的ACI患者,根据HT严重程度等级化分1级、2级、3级、4级,各严重程度患者分别为27例、22例、18例、13例。选取50例在接受非溶栓治疗后未出现HT的ACI患者作为对照(非转化组)。比较各组的血清NLR、PLR、hs-CRP、Hb、D-D表达水平差异,并通过Spearman秩相关检验分析NLR、PLR、hs-CRP、Hb、D-D与HT严重程度的相关性。并采用ROC曲线分析NLR、PLR、hsCRP、Hb、D-D对ACI患者HT的预测作用。结果 转化组的血清NLR、PLR、hs-CRP、D-D表达水平均高于非转化组,Hb的表达水平低于非转化组(均P<0.05)。1级、2级、3级、4级患者的血清NLR、PLR、hs-CRP、D-D表达水平依次上升,Hb表达水平依次降低,差异有统计学意义(P<0.05),且组间两两比较,差异有统计学意义(P<0.05)。ACI非溶栓治疗后HT严重程度与血清NLR、PLR、hs-CRP、D-D均呈正相关(r值分别为:0.729、0.692、0.601、0.578,均P<0.05),与Hb呈负相关(r值为:-0.678,P<0.05)。ROC曲线分析显示,NLR的AUC值为0.896[95%CI(0.737~0.962)]、PLR的AUC值为0.775 [95%CI(0.626~0.917)]、hs-CRP的AUC值为0.840 [95%CI(0.712~0.948)]、Hb的AUC值为0.782 [95%CI(0.493~0.914)]、D-D的AUC值为0.725 [95%CI(0.403~0.852)]。结论 ACI患者的NLR值、PLR值、hs-CRP、Hb、D-D表达水平与其在非溶栓治疗后发生HT存在相关性,临床上可将这些指标作为预测HT风险。

基于3Dslope的露天矿边坡稳定性计算

针对F3断层切割平庄西露天矿西北帮边坡形成三维滑体,采用二维计算方法已经不能满足计算要求,因此采用3Dslope边坡计算软件对西北帮边坡进行稳定性计算.结果表明:通过分析西北帮边坡的影响因素,确定潜在滑坡模式,计算由F3断层和弱层以及假设的侧滑面控制的三维滑体边坡稳定系数.从而确定滑体的大小,控制F3断层影响范围,采取相应的治理措施,该成果为安全生产提供技术支持.

基于Transformer的融合信息增强3D目标检测算法

针对当前3D目标检测算法将不同模态数据融合时会产生错位现象,从而破坏数据之间的关联性并造成数据损失的问题,提出了一种基于Transformer的融合信息增强3D目标检测算法。首先设计了Transformer双域融合特征区域建议模块,利用变形注意力机制,将提取到的雷达点云特征和图像特征进行双域特征融合,用于生成3D预选框;其次,通过设计的深度补全机制的特征信息增强模块,补全密集的深度和特征语义信息来完成框的细化;最后,设计了多模态特征交叉注意力模块,采用动态交叉注意力机制来获得不同模态间的相关性,从而将特征信息有效对齐融合。在Kitti、Nuscences和Waymo数据集上的实验结果证明了该算法的有效性和通用性。大量的消融实验证明了该算法各个模块的有效性。在实车平台上的实验结果表明,该算法在复杂的实际环境中具有优秀的鲁棒性。

基于Vision Transformer的时空卷积网络设计

目前主流人体动作识别大部分都是基于卷积神经网络(Convolutional Neural Network,CNN)实现,而CNN容易忽略视频中的空间位置信息,从而降低了视频空间频域中动作识别能力。同时传统CNN不能快速定位到关键的特征位置,并且在训练过程中不能并行计算导致效率低。为了解决传统CNN在处理时间频域和多并行计算问题,提出了基于视觉Transformer(Vision Transformer,ViT)和3D卷积网络学习时空特征(Learning Spatiotemporal Features with 3D Convolutional Network,C3D)的人体动作识别算法。使用C3D提取视频的多维特征图、ViT的特征切片窗口对多维特征进行全局特征分割;使用Transformer的编码-解码模块对视频中人体动作进行预测。实验结果表明,所提的人体动作识别算法在UCF-101、HMDB51数据集上提高了动作识别的准确率。

Nonlinear Dispersion Effect on Wave Transformation

A new nonlinear dispersion relation is given in this paper, which can overcome the limitation of the intermediate minimum value in the dispersion relation proposed by Kirby and Dalrymple (1986). and which has a better approximation to Hedges' empirical relation than the modified relations by Hedges (1987). Kirby and Dalrymple (1987) for shallow waters. The new dispersion relation is simple in form, thus it can be used easily in practice. Meanwhile, a general explicit approximation to the new dispersion and other and other nonlinear dispersion relations is given. By use of the explicit approximation to the new dispersion relation along with the mild slope equation taking into account weakly nonlinenr effect, a mathematical model is obtained, and it is applied to laboratory data. The results show that the model developed with the new dispersion relation predicts wave transformation over complicated topography quite well.

Q-factor estimation in CMP gather and the continuous spectral ratio slope method

The attenuation factor or quality factor(Q-factor or Q) has been used to measure the energy attenuation of seismic waves propagating in underground media. Many methods are used to estimate the Q-factor. We propose a method to calculate the Q-factor based on the prestack Q-factor inversion and the generalized S-transform. The proposed method specifies a standard primary wavelet and calculates the cumulative Q-factors; then, it finds the interlaminar Q-factors using the relation between Q and offset(QVO) and the Dix formula. The proposed method is alternative to methods that calculate interlaminar Q-factors after horizon picking. Because the frequency spectrum of each horizon can be extracted continuously on a 2D time–frequency spectrum, the method is called the continuous spectral ratio slope(CSRS) method. Compared with the other Q-inversion methods, the method offers nearly effortless computations and stability, and has mathematical and physical significance. We use numerical modeling to verify the feasibility of the method and apply it to real data from an oilfield in Ahdeb, Iraq. The results suggest that the resolution and spatial stability of the Q-profile are optimal and contain abundant interlaminar information that is extremely helpful in making lithology and fluid predictions.

Nonlinear Dispersion Relation in Wave Transformation

A nonlinear dispersion relation is presented to model the nonlinear dispersion of waves over the whole range of possible water depths. It reduces the phase speed over prediction of both Hedges′ modified relation and Kirby and Dalrymple′s modified relation in the region of 1< kh <1 5 for small wave steepness and maintains the monotonicity in phase speed variation for large wave steepness. And it has a simple form. By use of the new nonlinear dispersion relation along with the mild slope equation taking into account weak nonlinearity, a mathematical model of wave transformation is developed and applied to laboratory data. The results show that the model with the new dispersion relation can predict wave transformation over complicated bathymetry satisfactorily.

Methods applied in Australian industry to evaluate coal mine slope stability

Strata failure is a principal hazard in open cut coal mining as it has the potential to cause multiple fatalities.Prior to the excavation of any slope,a geotechnical assessment should review the likely slope performance,including the risk of slope failure.Controls to manage this risk to an acceptable level should accompany the geotechnical analysis.A survey of 43 practising geotechnical engineers indicated that kinematic and 2D limit equilibrium analyses were the methods most commonly applied to analyse excavated slope stability.While these methods are well established and widely applied in the broad rock engineering disciplines(e.g.civil,hard rock),a recent review of over 60 slope failures suggests these methods have limited suitability for modelling the complex failure mechanisms observed in excavated coal mine slopes.Kinematic techniques do not adequately capture the rock mass component of excavated slope failure and do not provide a geospatial location of potential failure,while 2D limit equilibrium techniques do not adequately capture the 3D mechanisms of excavated slope failures.Methods which do consider the 3D mechanisms of slope failure are summarized for industry consideration and application.

Application of Three-Dimensional Laser Scanning and Surveying in Geological Investigation of High Rock Slope

The appearance of 3D laser scanning technology is one of the most important technology revolutions in surveying and mapping field. It can be widely used in many interrelated fields, such as engineering constructions and 3D measurements, owing to its prominent characteristics of the high efficiency and high precision. At present its application is still in the initial state, and it is quite rarely used in China, especially in geotechnical engineering and geological engineering fields. Starting with a general introduction of 3D laser scanning technology, this article studies how to apply the technology to high rock slope investigations. By way of a case study, principles and methods of quick slope documentation and occurrence measurement of discontinuities are discussed and analyzed. Analysis results show that the application of 3D laser scanning technology to geotechnical and geological engineering has a great prospect and value.

Monitoring slope deformation using a 3-D laser image scanning system: a case study

An ILRIS-36D 3-D laser image scanning system was used to monitor the Anjialing strip mine slope on Pingshuo in Shanxi province. The basic working principles, performance indexes, features and data collection and processing methods are illus-trated. The point cloud results are analyzed in detail. The rescale range analysis method was used to analyze the deformation char-acteristics of the slope. The results show that the trend of slope displacement is stable and that the degree of landslide danger is low. This work indicates that 3-D laser image scanning can supply multi-parameter, high precision real time data over long distances. These data can be used to study the distortion of the slope quickly and accurately.