引入自上向下特征融合的小目标检测算法

针对原始SSD(Single Shot Multibox Detector)算法未充分利用各特征层之间关系导致浅层特征层缺乏小目标语义信息的问题,为了提高对小目标的检测能力,提出了一种结合PANet多尺度特征融合网络和自上向下特征融合路径的TTB-SSD(Top to Bottom SSD)改进算法。首先,使用PANet多尺度特征融合网络对特征进行反复提取,从而获得丰富的多尺度语义信息;然后,使用一种深层特征融合模块将浅层特征层的空间信息传递到深层特征层,进而更准确地对小目标进行定位;最后,为了增强浅层特征层的语义信息,构造了自上向下的特征融合路径,从而强化浅层对小目标检测的准确率。实验结果表明,在PASCAL VOC2007测试集检测的mAP(Mean Average Precision)值达到80.5%,对目标的mAP较原始SSD提高了5.7%,证明了该算法对小目标检测的有效性。     

基于改进的SSD监理目标检测研究

针对装饰装修工程中由人工验收带来的诸多问题,文中提出了一种改进的SSD算法并将其应用于监理工作来代替人工验收,推动智能监理的实现。由于SSD算法存在对同一目标复检以及小目标检测效果欠佳等问题,故文中利用DPN网络替换基础特征提取网络VGG16。DPN结合了Resnet和Densenet的优点,具有更好的特征提取能力。通过加权FPN融合特征图,突出不同层特征图的贡献,丰富用于预测的特征图语义。利用深度可分离卷积降低模型的参数量,提高算法的推理速度。实验对比发现,改进后模型的平均精度提升了3.47%,对小数目检测平均精度的提升可达15%,证明新模型在监理目标检测任务中效果良好。     

Synthesis and thermal decomposition of 3-aroyl-4-aryl-2-pyrazolines

Summary 3-Aroyl-4-aryl-2-pyrazolines (21–40) have been synthesized by the reaction of ,-unsaturated ketones (1–20) with diazomethane. These 2-pyrazolines gave -methyl-,-unsaturated ketones (41–46) on thermal denitrogenation.Dedicated to Prof. Dr.F. Sauter on the occasion of his 65^th birthday     

Quo Vadis total reflection X-ray fluorescence?

The multielement trace analytical method ‘total reflection X-ray fluorescence’ (TXRF) has become a successfully established method in the semiconductor industry, particularly, in the ultra trace element analysis of silicon wafer surfaces. TXRF applications can fulfill general industrial requirements on daily routine of monitoring wafer cleanliness up to 300 mm diameter under cleanroom conditions. Nowadays, TXRF and hyphenated TXRF methods such as ‘vapor phase decomposition (VPD)-TXRF’, i.e. TXRF with a preceding surface and acid digestion and preconcentration procedure, are automated routine techniques (‘wafer surface preparation system’, WSPS). A linear range from 10⁸ to 10¹⁴ [atoms/cm²] for some elements is regularly controlled. Instrument uptime is higher than 90%. The method is not tedious and can automatically be operated for 24 h/7 days. Elements such as S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Br, Sn, Sb, Ba and Pb are included in the software for standard peak search. The detection limits of recovered elements are between 1×10¹¹ and 1×10⁷ [atoms/cm²] depending upon X-ray excitation energy and the element of interest. For the determination of low Z elements, i.e. Na, Al and Mg, TXRF has also been extended but its implementation for routine analysis needs further research. At present, VPD-TXRF determination of light elements is viable in a range of 10⁹ [atoms/cm²]. Novel detectors such as silicon drift detectors (SDD) with an active area of 5 mm², 10 mm² or 20 mm², respectively, and multi-array detectors forming up to 70 mm² are commercially available. The first SDD with 100 mm² (!) area and integrated backside FET is working under laboratory conditions. Applications of and comparison with ICP-MS, HR-ICP-MS and SR-TXRF, an extension of TXRF capabilities with an extremely powerful energy source, are also reported.     

基于MEMS传感器的无线位置跟踪控制仿真应用

针对嵌入式工业控制系统终端显示的需要,提出了一种基于微处理器R1610和LCD控制器SSD1963的液晶显示模块的设计方法。详细阐述了硬件接口电路的设计和控制软件的编程,重点介绍了R1610、SSD1963以及TFT液晶屏AT070TN83之间的硬件连接方法,分析了SSD1963液晶控制的使用方法以及针对液晶屏显示时序的系统设置。结合显示控制实例实现了图像在液晶屏上的显示,显示效果良好。该显示模块的设计具有一定的应用前景和参考价值。     

基于LPC2210和SSD1339的PM-OLED驱动控制电路

为了避免早期无源矩阵有机电致发光器件驱动控制电路的一些缺陷,如:"串扰"和"交叉"效应以及电路连接比较复杂,用两种方法实现了基于飞利浦公司生产的LPC2210控制芯片和晶门科技公司的SSD1339驱动芯片的驱动控制电路。首先介绍了有机电致发光器件的结构和发光原理以及芯片SSD1339和LPC2210的主要的特点;分析和比较了SSD1339的8080系列并行口和LPC2210外部存储控制单元的读写时序;分别利用LPC2210的通用输入输出单元和外部存储控制单元,成功的控制SSD1339驱动128RGB×128点阵有机电致发光屏。实验结果表明:两种方法不仅可以有效地克服早期驱动控制电路的缺陷,而且可以使有机电致发光屏显示出高质量的图片;用外部存储控制单元实现的驱动控制电路,可以实现约80 Hz的驱动帧频;而使用通用输入输出单元实现的驱动控制电路,可以单步跟踪数据的传输,因此它具有方便查错的优势。本次实验为在不同的集成环境下无源矩阵有机电致发光器件驱动控制电路的设计奠定基础。     

基于MobileNet-SSD的红外人脸检测算法

针对"可见光"+"热红外"融合的双光人脸检测算法对硬件依赖性高的劣势.本文提出一种基于MobileNet-SSD的红外人脸检测算法,该算法可以直接检测出红外图像中的人脸区域,对硬件的依赖性较低.同时实验结果表明该算法的检测精度和实时性均有所提高,可以直接运用到"疫情"期间的智能体温监控系统中.     

SSD技术及应用

详细介绍了SSD的技术现状以及传统硬盘与SSD的比较,对主要性能指标进行了说明:分析了影响SSD技术发展的五个方面:Flash、主控、Firmware部分、Software方面、操作系统方面;阐述了SSD的应用行业和未来发展.     

CAWR: Buffer Replacement with Channel‐Aware Write Reordering Mechanism for SSDs

A typical solid‐state drive contains several independent channels that can be operated in parallel. To exploit this channel‐level parallelism, a variety of works proposed to split consecutive write sequences into small segments and schedule them to different channels. This scheme exploits the parallelism but breaks the spatial locality of write traffic; thus, it is able to significantly degrade the efficiency of garbage collection. This paper proposes a channel‐aware write reordering (CAWR) mechanism to schedule write requests to different channels more intelligently. The novel mechanism encapsulates correlated pages into a cluster beforehand. All pages belonging to a cluster are scheduled to the same channels to exploit spatial locality, while different clusters are scheduled to different channels to exploit the parallelism. As CAWR covers both garbage collection and I/O performance, it outperforms existing schemes significantly. Trace‐driven simulation results demonstrate that the CAWR mechanism reduces the average response time by 26% on average and decreases the valid page copies by 10% on average, while achieving a similar hit ratio to that of existing mechanisms.     

SFE-SSD: 浅层特征增强的SSD算法在小目标检测问题中的研究与应用

SSD (Single Shot Multibox Dectetor)算法由于具有高速且高精度的检测性能,是目前最好的目标检测算法之一.但由于提取检测框的特征层的特征信息不足, SSD算法在小目标检测任务中表现不佳.为了解决这个问题,目前大部分方法以严重牺牲检测速度为代价提升目标检测模型的精度. 本文提出了SFE-SSD (Shallow Feature Enhancement SSD)提升SSD模型在小目标检测任务中的性能.首先我们采用反卷积操作对SSD算法中检测框金字塔特征层的最浅特征层进行特征扩张.接着通过特征融合机制对扩张后的特征层进行特征增强操作.浅层特征增强策略与SSD 的原始框提取金字塔特征层是并行结构,一定程度上是可以减少检测速度的损失.实验结果显示,我们的方法在PASCAL VOC 2007数据库上精度达到了78.4\%mAP高于SSD算法1.2\%,检测速度达到了81帧/秒,并且在小目标检测任务中有着显著的提升.