UFLD(ultra fast structure aware deep lane detection)是一种轻量化车道线检测模型,为提升模型的检测精度,对模型进行改进。引入CAM(channel attention mechanism)使模型能更关注携带重要车道线信息的特征通道和像素;为了感知车道线...UFLD(ultra fast structure aware deep lane detection)是一种轻量化车道线检测模型,为提升模型的检测精度,对模型进行改进。引入CAM(channel attention mechanism)使模型能更关注携带重要车道线信息的特征通道和像素;为了感知车道线的细节信息,引入ASPP(atrous spatial pyramid pooling)扩大卷积过程的感受野,提高模型分割精度;搭建引入CAM和ASPP后的改进模型,并在改进的模型上进行实验。实验结果表明:在TuSimple数据集上以ResNet18为主干网络的模型检测精度由95.81%提升至95.98%,以ResNet34为主干网络的模型检测精度由95.84%提升至96.12%;在CULane数据集上,无论是以ResNet18还是以ResNet34为主干网络模型,其平均精度均有不同程度的提高。展开更多
In convolutional neural networks,pooling methods are used to reduce both the size of the data and the number of parameters after the convolution of the models.These methods reduce the computational amount of convoluti...In convolutional neural networks,pooling methods are used to reduce both the size of the data and the number of parameters after the convolution of the models.These methods reduce the computational amount of convolutional neural networks,making the neural network more efficient.Maximum pooling,average pooling,and minimum pooling methods are generally used in convolutional neural networks.However,these pooling methods are not suitable for all datasets used in neural network applications.In this study,a new pooling approach to the literature is proposed to increase the efficiency and success rates of convolutional neural networks.This method,which we call MAM(Maximum Average Minimum)pooling,is more interactive than other traditional maximum pooling,average pooling,and minimum pooling methods and reduces data loss by calculating the more appropriate pixel value.The proposed MAM pooling method increases the performance of the neural network by calculating the optimal value during the training of convolutional neural networks.To determine the success accuracy of the proposed MAM pooling method and compare it with other traditional pooling methods,training was carried out on the LeNet-5 model using CIFAR-10,CIFAR-100,and MNIST datasets.According to the results obtained,the proposed MAM pooling method performed better than the maximum pooling,average pooling,and minimum pooling methods in all pool sizes on three different datasets.展开更多
Pooling,unpooling/specialization,and discretionary task completion are typical operational strategies in queueing systems that arise in healthcare,call centers,and online sales.These strategies may have advantages and...Pooling,unpooling/specialization,and discretionary task completion are typical operational strategies in queueing systems that arise in healthcare,call centers,and online sales.These strategies may have advantages and disadvantages in different operational environments.This paper uses the M/M/1 and M/M/2 queues to study the impact of pooling,specialization,and discretionary task completion on the average queue length.Closed-form solutions for the average M/M/2 queue length are derived.Computational examples illustrate how the average queue length changes with the strength of pooling,specialization,and discretionary task completion.Finally,several conjectures are made in the paper.展开更多
The state of in situ stress is a crucial parameter in subsurface engineering,especially for critical projects like nuclear waste repository.As one of the two ISRM suggested methods,the overcoring(OC)method is widely u...The state of in situ stress is a crucial parameter in subsurface engineering,especially for critical projects like nuclear waste repository.As one of the two ISRM suggested methods,the overcoring(OC)method is widely used to estimate the full stress tensors in rocks by independent regression analysis of the data from each OC test.However,such customary independent analysis of individual OC tests,known as no pooling,is liable to yield unreliable test-specific stress estimates due to various uncertainty sources involved in the OC method.To address this problem,a practical and no-cost solution is considered by incorporating into OC data analysis additional information implied within adjacent OC tests,which are usually available in OC measurement campaigns.Hence,this paper presents a Bayesian partial pooling(hierarchical)model for combined analysis of adjacent OC tests.We performed five case studies using OC test data made at a nuclear waste repository research site of Sweden.The results demonstrate that partial pooling of adjacent OC tests indeed allows borrowing of information across adjacent tests,and yields improved stress tensor estimates with reduced uncertainties simultaneously for all individual tests than they are independently analysed as no pooling,particularly for those unreliable no pooling stress estimates.A further model comparison shows that the partial pooling model also gives better predictive performance,and thus confirms that the information borrowed across adjacent OC tests is relevant and effective.展开更多
针对目前遥感影像中烤烟边缘识别效率低且识别精度低等问题,文章提出一种融合轻量化ASPP和U-Net框架的遥感影像烤烟种植区域提取模型。首先,该模型在U-Net编码层和解码层连接处加入轻量化空洞空间金字塔池化模块;其次,该模型将线性整流...针对目前遥感影像中烤烟边缘识别效率低且识别精度低等问题,文章提出一种融合轻量化ASPP和U-Net框架的遥感影像烤烟种植区域提取模型。首先,该模型在U-Net编码层和解码层连接处加入轻量化空洞空间金字塔池化模块;其次,该模型将线性整流函数(Rectified Linear Unit,ReLU)替换为ReLU6激活函数,能够在低精度计算时压缩动态范围,从而使算法更具鲁棒性;最后,该模型通过采用形态学孔洞填充构建标签图后处理算法,实现分割结果优化。为验证该模型框架的有效性和适用性,文章采用无人机遥感影像作为实验数据集,构建与传统语义分割模型的对比实验以及消融实验等。实验结果表明,通过与FCN、U-Net、SegNet和DeepLabV3+等传统语义分割算法相比较,文章提出的模型获得了较好的分割效果,其像素准确率和平均交并比分别为93.7%和84.1%。此外,该模型在保证模型精度的情况下,还能够提高模型的计算速度。展开更多
In order to enhance the performance of the CNN-based segmentation models for bone metastases, this study proposes a segmentation method that integrates dual-pooling, DAC, and RMP modules. The network consists of disti...In order to enhance the performance of the CNN-based segmentation models for bone metastases, this study proposes a segmentation method that integrates dual-pooling, DAC, and RMP modules. The network consists of distinct feature encoding and decoding stages, with dual-pooling modules employed in encoding stages to maintain the background information needed for bone scintigrams diagnosis. Both the DAC and RMP modules are utilized in the bottleneck layer to address the multi-scale problem of metastatic lesions. Experimental evaluations on 306 clinical SPECT data have demonstrated that the proposed method showcases a substantial improvement in both DSC and Recall scores by 3.28% and 6.55% compared the baseline. Exhaustive case studies illustrate the superiority of the methodology.展开更多
作为高铁牵引供电系统的重要组成部分,接触网系统承担着向动车组传输电能的重要功能.实际工程运营表明,受弓网交互产生的持续冲击以及外部环境的影响,接触网支撑部件可能会出现“松、脱、断、裂”等缺陷,导致接触网结构可靠性下降,严重...作为高铁牵引供电系统的重要组成部分,接触网系统承担着向动车组传输电能的重要功能.实际工程运营表明,受弓网交互产生的持续冲击以及外部环境的影响,接触网支撑部件可能会出现“松、脱、断、裂”等缺陷,导致接触网结构可靠性下降,严重影响接触网系统稳定运行.因此,及时精确定位接触网支撑部件(CSCs),对保障高铁安全运行和完善接触网检修维护策略具有重大意义.然而,CSCs的检测通常面临着零部件种类多、尺度差异大、部分零部件微小的问题.针对以上问题,本文提出一种基于多尺度融合金字塔焦点网络的接触网零部件检测算法,将平衡模块和特征金字塔模块相结合,提高对小目标的检测性能.首先,设计了可分离残差金字塔聚合模块(SRPAM),用于优化模型多尺度特征提取能力、扩大感受野,缓解CSCs检测的多尺度问题;其次,设计了一种基于平衡特征金字塔的路径聚合网络(PA-BFPN),用于提升跨层特征融合效率和小目标检测性能.最后,通过对比试验、可视化实验和消融实验证明了所提方法的有效性和优越性.其中,所提的MFPFCOS在CSCs数据集上的检测精度(mAP)能够在达到48.6%的同时,实现30的FLOPs(Floating point operations per second),表明所提方法能够在检测精度和检测速度之间保持良好的平衡.展开更多
池式研究型反应堆是全球多用途研究堆中的重点类型,因其在安全特性、多用途性和运行维护等方面的突出表现而备受关注。在池式堆中,热功率为1~10 MW的小型研究堆方案设计最为成熟,应用场景最为广泛。为了探索未来池式研究堆堆芯设计及其...池式研究型反应堆是全球多用途研究堆中的重点类型,因其在安全特性、多用途性和运行维护等方面的突出表现而备受关注。在池式堆中,热功率为1~10 MW的小型研究堆方案设计最为成熟,应用场景最为广泛。为了探索未来池式研究堆堆芯设计及其应用场景的发展趋势,首先,根据全球开展RERTR(Reduced Enrichment Research and Test Reactor)低浓化项目并进行堆芯重新设计的小型池式研究堆不同的堆芯方案进行对比,研究分析未来小型池式研究堆堆芯可采用的燃料类型和组件结构,以及目前全球小型池式研究堆的应用情况。其次,总结了小型池式研究堆在燃料类型和堆芯结构两个方面的发展现状,汇总了研究堆各类中子应用场景的技术指标。最终,通过横向对比探究分析推判:未来小型池式研究堆将采用紧凑型堆芯设计,采用高密度的低浓缩铀燃料,以紧凑可移动式小堆芯为基础,以大水池内中子源应用设施为主要发展方向。展开更多
文摘UFLD(ultra fast structure aware deep lane detection)是一种轻量化车道线检测模型,为提升模型的检测精度,对模型进行改进。引入CAM(channel attention mechanism)使模型能更关注携带重要车道线信息的特征通道和像素;为了感知车道线的细节信息,引入ASPP(atrous spatial pyramid pooling)扩大卷积过程的感受野,提高模型分割精度;搭建引入CAM和ASPP后的改进模型,并在改进的模型上进行实验。实验结果表明:在TuSimple数据集上以ResNet18为主干网络的模型检测精度由95.81%提升至95.98%,以ResNet34为主干网络的模型检测精度由95.84%提升至96.12%;在CULane数据集上,无论是以ResNet18还是以ResNet34为主干网络模型,其平均精度均有不同程度的提高。
文摘In convolutional neural networks,pooling methods are used to reduce both the size of the data and the number of parameters after the convolution of the models.These methods reduce the computational amount of convolutional neural networks,making the neural network more efficient.Maximum pooling,average pooling,and minimum pooling methods are generally used in convolutional neural networks.However,these pooling methods are not suitable for all datasets used in neural network applications.In this study,a new pooling approach to the literature is proposed to increase the efficiency and success rates of convolutional neural networks.This method,which we call MAM(Maximum Average Minimum)pooling,is more interactive than other traditional maximum pooling,average pooling,and minimum pooling methods and reduces data loss by calculating the more appropriate pixel value.The proposed MAM pooling method increases the performance of the neural network by calculating the optimal value during the training of convolutional neural networks.To determine the success accuracy of the proposed MAM pooling method and compare it with other traditional pooling methods,training was carried out on the LeNet-5 model using CIFAR-10,CIFAR-100,and MNIST datasets.According to the results obtained,the proposed MAM pooling method performed better than the maximum pooling,average pooling,and minimum pooling methods in all pool sizes on three different datasets.
文摘Pooling,unpooling/specialization,and discretionary task completion are typical operational strategies in queueing systems that arise in healthcare,call centers,and online sales.These strategies may have advantages and disadvantages in different operational environments.This paper uses the M/M/1 and M/M/2 queues to study the impact of pooling,specialization,and discretionary task completion on the average queue length.Closed-form solutions for the average M/M/2 queue length are derived.Computational examples illustrate how the average queue length changes with the strength of pooling,specialization,and discretionary task completion.Finally,several conjectures are made in the paper.
基金supported by the Guangdong Basic and Applied Basic Research Foundation(2023A1515011244).
文摘The state of in situ stress is a crucial parameter in subsurface engineering,especially for critical projects like nuclear waste repository.As one of the two ISRM suggested methods,the overcoring(OC)method is widely used to estimate the full stress tensors in rocks by independent regression analysis of the data from each OC test.However,such customary independent analysis of individual OC tests,known as no pooling,is liable to yield unreliable test-specific stress estimates due to various uncertainty sources involved in the OC method.To address this problem,a practical and no-cost solution is considered by incorporating into OC data analysis additional information implied within adjacent OC tests,which are usually available in OC measurement campaigns.Hence,this paper presents a Bayesian partial pooling(hierarchical)model for combined analysis of adjacent OC tests.We performed five case studies using OC test data made at a nuclear waste repository research site of Sweden.The results demonstrate that partial pooling of adjacent OC tests indeed allows borrowing of information across adjacent tests,and yields improved stress tensor estimates with reduced uncertainties simultaneously for all individual tests than they are independently analysed as no pooling,particularly for those unreliable no pooling stress estimates.A further model comparison shows that the partial pooling model also gives better predictive performance,and thus confirms that the information borrowed across adjacent OC tests is relevant and effective.
文摘针对目前遥感影像中烤烟边缘识别效率低且识别精度低等问题,文章提出一种融合轻量化ASPP和U-Net框架的遥感影像烤烟种植区域提取模型。首先,该模型在U-Net编码层和解码层连接处加入轻量化空洞空间金字塔池化模块;其次,该模型将线性整流函数(Rectified Linear Unit,ReLU)替换为ReLU6激活函数,能够在低精度计算时压缩动态范围,从而使算法更具鲁棒性;最后,该模型通过采用形态学孔洞填充构建标签图后处理算法,实现分割结果优化。为验证该模型框架的有效性和适用性,文章采用无人机遥感影像作为实验数据集,构建与传统语义分割模型的对比实验以及消融实验等。实验结果表明,通过与FCN、U-Net、SegNet和DeepLabV3+等传统语义分割算法相比较,文章提出的模型获得了较好的分割效果,其像素准确率和平均交并比分别为93.7%和84.1%。此外,该模型在保证模型精度的情况下,还能够提高模型的计算速度。
文摘In order to enhance the performance of the CNN-based segmentation models for bone metastases, this study proposes a segmentation method that integrates dual-pooling, DAC, and RMP modules. The network consists of distinct feature encoding and decoding stages, with dual-pooling modules employed in encoding stages to maintain the background information needed for bone scintigrams diagnosis. Both the DAC and RMP modules are utilized in the bottleneck layer to address the multi-scale problem of metastatic lesions. Experimental evaluations on 306 clinical SPECT data have demonstrated that the proposed method showcases a substantial improvement in both DSC and Recall scores by 3.28% and 6.55% compared the baseline. Exhaustive case studies illustrate the superiority of the methodology.
文摘作为高铁牵引供电系统的重要组成部分,接触网系统承担着向动车组传输电能的重要功能.实际工程运营表明,受弓网交互产生的持续冲击以及外部环境的影响,接触网支撑部件可能会出现“松、脱、断、裂”等缺陷,导致接触网结构可靠性下降,严重影响接触网系统稳定运行.因此,及时精确定位接触网支撑部件(CSCs),对保障高铁安全运行和完善接触网检修维护策略具有重大意义.然而,CSCs的检测通常面临着零部件种类多、尺度差异大、部分零部件微小的问题.针对以上问题,本文提出一种基于多尺度融合金字塔焦点网络的接触网零部件检测算法,将平衡模块和特征金字塔模块相结合,提高对小目标的检测性能.首先,设计了可分离残差金字塔聚合模块(SRPAM),用于优化模型多尺度特征提取能力、扩大感受野,缓解CSCs检测的多尺度问题;其次,设计了一种基于平衡特征金字塔的路径聚合网络(PA-BFPN),用于提升跨层特征融合效率和小目标检测性能.最后,通过对比试验、可视化实验和消融实验证明了所提方法的有效性和优越性.其中,所提的MFPFCOS在CSCs数据集上的检测精度(mAP)能够在达到48.6%的同时,实现30的FLOPs(Floating point operations per second),表明所提方法能够在检测精度和检测速度之间保持良好的平衡.
文摘池式研究型反应堆是全球多用途研究堆中的重点类型,因其在安全特性、多用途性和运行维护等方面的突出表现而备受关注。在池式堆中,热功率为1~10 MW的小型研究堆方案设计最为成熟,应用场景最为广泛。为了探索未来池式研究堆堆芯设计及其应用场景的发展趋势,首先,根据全球开展RERTR(Reduced Enrichment Research and Test Reactor)低浓化项目并进行堆芯重新设计的小型池式研究堆不同的堆芯方案进行对比,研究分析未来小型池式研究堆堆芯可采用的燃料类型和组件结构,以及目前全球小型池式研究堆的应用情况。其次,总结了小型池式研究堆在燃料类型和堆芯结构两个方面的发展现状,汇总了研究堆各类中子应用场景的技术指标。最终,通过横向对比探究分析推判:未来小型池式研究堆将采用紧凑型堆芯设计,采用高密度的低浓缩铀燃料,以紧凑可移动式小堆芯为基础,以大水池内中子源应用设施为主要发展方向。
