微量热法研究黄嘌呤氧化酶反应

在热导工热量计双参数理论模型的基础上，建立了较快酶仲反应研究的双参数初始速度法的热动力学模型，用微量热法研究了黄嘌呤氧化酶催化氧化黄嘌呤的热动力学，该较快酶促反应遵循Ｍｉｃｈａｅｌｉｓ－Ｍｅｎｔｅｎ这，在２９８．１５Ｋ和ＰＨ＝７．５时，其米氏常数为１．０４×１０＾－３ｍｏｌ．Ｌ＾－１，与文献结果相符。     

Monolithic silica rod columns for high-efficiency reversed-phase liquid chromatography

Chromatographic properties of a new type of monolithic silica rod columns were examined. Silica rod columns employed for the study were prepared from tetramethoxysilane, modified with octadecylsilyl moieties, and encased in a stainless-steel protective column with two polymer layers between the silica and the stainless-steel tubing. A 25 cm column provided up to 45,000 theoretical plates for aromatic hydrocarbons, or a minimum plate height of about 5.5 μm, at optimum linear velocity of ca. 2.3 mm/s and back pressure of 7.5 MPa in an acetonitrile-water (80/20, v/v) mobile phase at 40°C. The permeability of the column was similar to that of a column packed with 5 μm particles, with K(F) about 2.4×10(-14) m(2) (based on the superficial linear velocity of the mobile phase), while the plate height value equivalent to that of a column packed with 2.5 μm particles. Generation of 80,000-120,000 theoretical plates was feasible with back pressure below 30 MPa by employing two or three 25 cm columns connected in series. The use of the long columns enabled facile generation of large numbers of theoretical plates in comparison with conventional monolithic silica columns or particulate columns. Kinetic plot analysis indicates that the monolithic columns operated at 30 MPa can provide faster separations than a column packed with totally porous 3-μm particles operated at 40 MPa in a range where the number of theoretical plates (N) is greater than 50,000.     

Faster analysis by reversed-phase high-performance liquid chromatography

Summary Many analysts are not taking full advantage of the high speed possibilities of modern LC. Some analytical procedures reported in the literature, and many in regular use in control laboratories, could be achieved in less time without loss in precision. Some factors which affect retention times are discussed and the advantages and disadvantages of employing shorter column lengths and finer packing materials in reversed-phase HPLC are examined. The effect on efficiency of increased flow rates with 10,5 and 3 m ODS materials is shown. The ability to couple shorter column lengths without loss of efficiency is also demonstrated. This allows a minimum length to be selected that gives adequate resolution. Examples of high speed separations are shown and limitations in state of the art HPLC equipment and chromatographic data systems are discussed briefly.Presented at the 14th International Symposium on Chromatography London, September, 1982     

Ungulate Detection and Species Classification from Camera Trap Images Using RetinaNet and Faster R-CNN

Changes in the ungulate population density in the wild has impacts on both the wildlife and human society. In order to control the ungulate population movement, monitoring systems such as camera trap networks have been implemented in a non-invasive setup. However, such systems produce a large number of images as the output, hence making it very resource consuming to manually detect the animals. In this paper, we present a new dataset of wild ungulates which was collected in Latvia. Moreover, we demonstrate two methods, which use RetinaNet and Faster R-CNN as backbones, respectively, to detect the animals in the images. We discuss the optimization of training and impact of data augmentation on the performance. Finally, we show the result of aforementioned tune networks over the real world data collected in Latvia.     

An OpenCL-Based FPGA Accelerator for Faster R-CNN

In recent years, convolutional neural network (CNN)-based object detection algorithms have made breakthroughs, and much of the research corresponds to hardware accelerator designs. Although many previous works have proposed efficient FPGA designs for one-stage detectors such as Yolo, there are still few accelerator designs for faster regions with CNN features (Faster R-CNN) algorithms. Moreover, CNN’s inherently high computational complexity and high memory complexity bring challenges to the design of efficient accelerators. This paper proposes a software-hardware co-design scheme based on OpenCL to implement a Faster R-CNN object detection algorithm on FPGA. First, we design an efficient, deep pipelined FPGA hardware accelerator that can implement Faster R-CNN algorithms for different backbone networks. Then, an optimized hardware-aware software algorithm was proposed, including fixed-point quantization, layer fusion, and a multi-batch Regions of interest (RoIs) detector. Finally, we present an end-to-end design space exploration scheme to comprehensively evaluate the performance and resource utilization of the proposed accelerator. Experimental results show that the proposed design achieves a peak throughput of 846.9 GOP/s at the working frequency of 172 MHz. Compared with the state-of-the-art Faster R-CNN accelerator and the one-stage YOLO accelerator, our method achieves 10× and 2.1× inference throughput improvements, respectively.     

基于改进Faster RCNN的零件识别方法研究北大核心CSCD

针对传统机器视觉检测识别方法受到装配零件之间相互遮挡、零件不同位姿、外部光照强度、小目标漏检影响,检测准确率不高的问题,提出了一种改进的Faster RCNN(region-based convolutional neural networks)零件识别方法。首先使用提取特征更好的ResNet101网络代替原始Faster RCNN模型中的VGG16特征提取网络;其次针对原始候选区域网络,增加2个新的锚点并重新设置候选框的纵横比,以得到15种尺寸不同的锚点;然后针对传统非极大值抑制(non-maximum suppression,NMS)方法因删除交并比大于阈值的候选框而出现漏检问题,使用Soft-NMS方法替换传统的NMS方法,从而减少密集区域漏检的情况;最后在训练模型阶段采用多尺度训练策略,降低漏检率,提高模型准确率。对零件的识别实验结果表明:改进后的Faster RCNN模型能够达到96.1%的精度,较原始模型提升了4.6%,可以满足光照较强、存在水渍干扰等较为复杂环境中零件的识别检测。