To track the nonlinear,non-Gaussian bearings-only maneuvering target accurately online,the constrained auxiliary particle filtering(CAPF)algorithm is presented.To restrict the samples into the feasible area,the soft m...To track the nonlinear,non-Gaussian bearings-only maneuvering target accurately online,the constrained auxiliary particle filtering(CAPF)algorithm is presented.To restrict the samples into the feasible area,the soft measurement constraints are implemented into the update routine via the1 regularization.Meanwhile,to enhance the sampling diversity and efficiency,the target kinetic features and the latest observations are involved into the evolution.To take advantage of the past and the current measurement information simultaneously,the sub-optimal importance distribution is constructed as a Gaussian mixture consisting of the original and modified priors with the fuzzy weighted factors.As a result,the corresponding weights are more evenly distributed,and the posterior distribution of interest is approximated well with a heavier tailor.Simulation results demonstrate the validity and superiority of the CAPF algorithm in terms of efficiency and robustness.展开更多
A laser-produced plasma(LPP) source was built using liquid as target and a Nd:YAG laser as the irradiation laser, and the LPP source's radiation with ethanol and acetone target respectively was measured by an AXUV...A laser-produced plasma(LPP) source was built using liquid as target and a Nd:YAG laser as the irradiation laser, and the LPP source's radiation with ethanol and acetone target respectively was measured by an AXUV100 silicon photodiode combined with a McPHERSON model 247 grazing incidence monochromator of the resolution Δλ≤0.075 nm and the wavelength scanning interval 0.5 nm. Both ethanol and acetone target LPP source had EUV emission at 11～20 nm wavelength. The comparison between the spectra of the two kinds of target materials shows that all the two kinds of target source's spectra are the result of oxygen ions' transitions under current source's parameters, but the spectrum intensity from different target sources is different. The spectra intensity from the ethanol target is higher than that from the acetone target. In addition, the target liquid is forced into the vacuum chamber by the background pressure supported by the connected external high pressure gas, and the influence of the background pressure on the source's intensity is investigated.展开更多
Soft tissue sarcoma are rare tumors arising mostly from embryonic mesoderm, that can affect almost any part of the human body, including the gastrointestinal tract. The prognosis associated with soft tissue sarcoma is...Soft tissue sarcoma are rare tumors arising mostly from embryonic mesoderm, that can affect almost any part of the human body, including the gastrointestinal tract. The prognosis associated with soft tissue sarcoma is still poor, mainly because of the low efficacy of traditional approaches based on surgery and chemotherapy. As a result of genetic and molecular analysis, several new target therapies have been developed, leading to a significant improvement in the survival of patients affected by advanced disease. In this review we aim to explore the therapeutic potential and benefit of target therapy in the management of gastrointestinal soft tissue sarcoma and the possible complications or pitfalls of such an approach.展开更多
针对交通目标检测时物体尺度变化大,检测目标较为密集的问题,基于YOLO(you only look once)v8s提出一种高效多尺度YOLO(fast multiscale powerful-YOLO,FMP-YOLO)模型。在主干网络中,引入基于部分卷积(partial convolution,PConv)与快...针对交通目标检测时物体尺度变化大,检测目标较为密集的问题,基于YOLO(you only look once)v8s提出一种高效多尺度YOLO(fast multiscale powerful-YOLO,FMP-YOLO)模型。在主干网络中,引入基于部分卷积(partial convolution,PConv)与快速傅里叶卷积(fast Fourier convolution,FFC)设计的Faster Block模块,减少了冗余计算和内存访问,提高了推理速度,改善了感受野受限的问题;在聚合网络中,使用改进后的组混洗卷积(group shuffle convolution,GSConv)替换普通卷积,更好地捕获不同尺度的特征,进一步降低了模型的参数量和计算量;将Powerful-IoU与软非极大值抑制(soft non maximum suppression,SoftNMS)结合替换原有的非极大值抑制算法(non maximum suppression,NMS),改善了由参数量降低导致特征学习能力下降的问题,并且提高了模型的精度。在SODA10M和MS COCO数据集上进行实验,实验结果表明,改进后的模型性能超越原始模型,参数量和计算量下降40%左右,mAP分别提高了1.7%和1.4%,FMP-YOLO在体积与精度上优于其他经典模型,具有较强的实用性。展开更多
在交通监控中进行车辆检测过程时,存在车辆互相遮挡和远距离目标尺寸不足的问题,导致在检测中存在漏检和误检情况。针对此问题,文中提出一种基于YOLOv4(You Only Look Once version 4)的多尺度融合与注意力机制的交通车辆检测算法。在YO...在交通监控中进行车辆检测过程时,存在车辆互相遮挡和远距离目标尺寸不足的问题,导致在检测中存在漏检和误检情况。针对此问题,文中提出一种基于YOLOv4(You Only Look Once version 4)的多尺度融合与注意力机制的交通车辆检测算法。在YOLOv4的路径聚合网络中增加一个新的特征层进行多尺度特征融合,提升模型对底层纹理特征的提取能力。在YOLO Head检测头前嵌入ECA(Efficient Channel Attention)通道注意力模块,对聚合后的特征进行合理的抑制和增强,将CIoU(Complete Intersection over Union)损失函数替换为Soft-CIoU损失函数,提高小目标车辆对损失函数的贡献度。在公开车辆数据集UA-DETRAC与KITTI中的实验结果表明,相较于原YOLOv4算法,所提算法的平均精度分别提升了2.45百分点和1.14百分点,检测速度达到41.67 frame·s^(-1)。相较于其他先进算法,所提算法在检测精度上表现良好。展开更多
基金supported by the National Natural Science Foundation of China(61773267)the Shenzhen Fundamental Research Project(JCYJ2017030214551952420170818102503604)
文摘To track the nonlinear,non-Gaussian bearings-only maneuvering target accurately online,the constrained auxiliary particle filtering(CAPF)algorithm is presented.To restrict the samples into the feasible area,the soft measurement constraints are implemented into the update routine via the1 regularization.Meanwhile,to enhance the sampling diversity and efficiency,the target kinetic features and the latest observations are involved into the evolution.To take advantage of the past and the current measurement information simultaneously,the sub-optimal importance distribution is constructed as a Gaussian mixture consisting of the original and modified priors with the fuzzy weighted factors.As a result,the corresponding weights are more evenly distributed,and the posterior distribution of interest is approximated well with a heavier tailor.Simulation results demonstrate the validity and superiority of the CAPF algorithm in terms of efficiency and robustness.
文摘A laser-produced plasma(LPP) source was built using liquid as target and a Nd:YAG laser as the irradiation laser, and the LPP source's radiation with ethanol and acetone target respectively was measured by an AXUV100 silicon photodiode combined with a McPHERSON model 247 grazing incidence monochromator of the resolution Δλ≤0.075 nm and the wavelength scanning interval 0.5 nm. Both ethanol and acetone target LPP source had EUV emission at 11～20 nm wavelength. The comparison between the spectra of the two kinds of target materials shows that all the two kinds of target source's spectra are the result of oxygen ions' transitions under current source's parameters, but the spectrum intensity from different target sources is different. The spectra intensity from the ethanol target is higher than that from the acetone target. In addition, the target liquid is forced into the vacuum chamber by the background pressure supported by the connected external high pressure gas, and the influence of the background pressure on the source's intensity is investigated.
文摘Soft tissue sarcoma are rare tumors arising mostly from embryonic mesoderm, that can affect almost any part of the human body, including the gastrointestinal tract. The prognosis associated with soft tissue sarcoma is still poor, mainly because of the low efficacy of traditional approaches based on surgery and chemotherapy. As a result of genetic and molecular analysis, several new target therapies have been developed, leading to a significant improvement in the survival of patients affected by advanced disease. In this review we aim to explore the therapeutic potential and benefit of target therapy in the management of gastrointestinal soft tissue sarcoma and the possible complications or pitfalls of such an approach.
文摘针对交通目标检测时物体尺度变化大,检测目标较为密集的问题,基于YOLO(you only look once)v8s提出一种高效多尺度YOLO(fast multiscale powerful-YOLO,FMP-YOLO)模型。在主干网络中,引入基于部分卷积(partial convolution,PConv)与快速傅里叶卷积(fast Fourier convolution,FFC)设计的Faster Block模块,减少了冗余计算和内存访问,提高了推理速度,改善了感受野受限的问题;在聚合网络中,使用改进后的组混洗卷积(group shuffle convolution,GSConv)替换普通卷积,更好地捕获不同尺度的特征,进一步降低了模型的参数量和计算量;将Powerful-IoU与软非极大值抑制(soft non maximum suppression,SoftNMS)结合替换原有的非极大值抑制算法(non maximum suppression,NMS),改善了由参数量降低导致特征学习能力下降的问题,并且提高了模型的精度。在SODA10M和MS COCO数据集上进行实验,实验结果表明,改进后的模型性能超越原始模型,参数量和计算量下降40%左右,mAP分别提高了1.7%和1.4%,FMP-YOLO在体积与精度上优于其他经典模型,具有较强的实用性。
文摘在交通监控中进行车辆检测过程时,存在车辆互相遮挡和远距离目标尺寸不足的问题,导致在检测中存在漏检和误检情况。针对此问题,文中提出一种基于YOLOv4(You Only Look Once version 4)的多尺度融合与注意力机制的交通车辆检测算法。在YOLOv4的路径聚合网络中增加一个新的特征层进行多尺度特征融合,提升模型对底层纹理特征的提取能力。在YOLO Head检测头前嵌入ECA(Efficient Channel Attention)通道注意力模块,对聚合后的特征进行合理的抑制和增强,将CIoU(Complete Intersection over Union)损失函数替换为Soft-CIoU损失函数,提高小目标车辆对损失函数的贡献度。在公开车辆数据集UA-DETRAC与KITTI中的实验结果表明,相较于原YOLOv4算法,所提算法的平均精度分别提升了2.45百分点和1.14百分点,检测速度达到41.67 frame·s^(-1)。相较于其他先进算法,所提算法在检测精度上表现良好。
