YOLSO: You Only Look Small Object

Small object detection is challenging and far from satisfactory. Most general object detectors suffer from two critical issues with small objects: (1) Feature extractor based on classification network cannot express the characteristics of small objects reasonably due to insufficient appearance information of targets and a large amount of background interference around them. (2) The detector requires a much higher location accuracy for small objects than for general objects. This paper proposes an effective and efficient small object detector YOLSO to address the above problems. For feature representation, we analyze the drawbacks in previous backbones and present a Half-Space Shortcut(HSSC) module to build a background-aware backbone. Furthermore, a coarse-to-fine Feature Pyramid Enhancement(FPE) module is introduced for layer-wise aggregation at a granular level to enhance the semantic discriminability. For loss function, we propose an exponential L1 loss to promote the convergence of regression, and a focal IOU loss to focus on prime samples with high classification confidence and high IOU. Both of them significantly improves the location accuracy of small objects. The proposed YOLSO sets state-of-the-art results on two typical small object datasets, MOCOD and VeDAI, at a speed of over 200 FPS. In the meantime, it also outperforms the baseline YOLOv3 by a wide margin on the common COCO dataset.     

Effects of high-pressure-modified soy 11S globulin on the gel properties and water-holding capacity of pork batter

The effects of high-pressure-modified soy 11S globulin (0.1, 200, and 400 MPa) on the gel properties, water-holding capacity, and water mobility of pork batter were investigated. The high-pressure-modified soy 11S globulin significantly increased (P < 0.05) the emulsion stability, cooking yield, hardness, springiness, chewiness, resilience, cohesiveness, the a^* and b^* values, and the G′ and G′′ values of pork batter at 80 °C, compared with those of 0.1 MPa-modified globulin. In contrast, the centrifugal loss and initial relaxation time of T_2b, T₂₁, and T₂₂ significantly decreased (P < 0.05). Meanwhile, the microstructure was denser, and the voids were smaller and more uniform compared with those of 0.1 MPa-modified globulin. In addition, the sample with 11S globulin modified at 400 MPa had the best water-holding capacity, gel structure, and gel properties among the samples. Overall, the use of high-pressure-modified soy 11S globulin improved the gel properties and water-holding capacity of pork batter, especially under 400 MPa.     

IRS辅助的SWIPT系统中基于能效优先的波束成形设计与优化

以智能反射面(intelligent reflecting surface,IRS)辅助的无线携能通信(simultaneous wireless information and power transfer,SWIPT)系统为背景,研究了该系统中基于能效优先的多天线发送端有源波束成形与IRS无源波束成形联合设计与优化方法。以最大化接收端的最小能效为优化目标,构造在发送端功率、接收端能量阈值、IRS相移等多约束下的非线性优化问题,用交替方向乘子法(alternating direction method of multipliers,ADMM)求解。采用Dinkelbach算法转化目标函数,通过奇异值分解(singular value decomposition,SVD)和半定松弛(semi-definite relaxation,SDR)得到发送端有源波束成形向量。采用SDR得到IRS相移矩阵与反射波束成形向量。结果表明,该系统显著降低了系统能量收集(energy harvesting,EH)接收端的能量阈值。当系统总电路功耗为?15 dBm时,所提方案的用户能效为300 KB/J。当IRS反射阵源数与发送天线数均为最大值时,系统可达最大能效。     

Preparation of high-purity SiC ceramics with good plasma corrosion resistance by hot-pressing sintering

High-purity SiC ceramic devices are applied in semiconductor industry owing to their outstanding properties. Nevertheless, it is difficult to densify SiC ceramics without any sintering additive even by HP sintering. In this work, high-purity and dense SiC ceramics were fabricated by HP sintering with very low amounts of sintering aids. Residual B content was only 556 ppm and relative density was more than 99.5%. Furthermore, thermal conductivity of as-prepared SiC ceramics was improved from 155 W m^?1 K^?1 to 167 W m^?1 K^?1 by increasing holding time and their plasma corrosion resistance was promoted in the meantime. The as-prepared high-purity SiC ceramics have broad application prospects in the field of semiconductor industry.     

Optical and electrical properties of ITO film on flexible fluorphlogopite substrate

Indium Tin Oxide (ITO) films were prepared, at room temperature, on a fluorphlogopite substrate using magnetron sputtering technology. At various temperatures of 500 °C, 600 °C, 700 °C, 800 °C, and 900 °C, the samples were (had) annealed for 2 h (a 2-h duration). The results showed improvement in the crystalline performance of ITO film at selected annealing temperatures, with a significant reduction in resistivity at 800 °C. The lowest resistivity is 4.08 × 10^?4 Ω-cm, which is nearly an order of magnitude lower than the unannealed sample. All samples have an average light transmittance above 85% in the visible light range (400–800 nm), and with increasing annealing temperature, the average light transmittance tends to decrease. Besides, at the sensitive wavelength of 550 nm, the light transmittance is as high as 93.74%. The sheet resistance testing of the sample was through the number of bending times, which revealed that with the increase of the number of bending, the sheet resistance increases. However, after 1200 bending times, the change rate of the sheet resistance remains below 5%. Thus, the ITO film prepared on the flexible fluorphlogopite substrate revealed excellent optical and electrical properties, good flexibility, and improved stability after high-temperature annealing, which guarantees successful application in flexible electronic devices.     

A robust model-free controller for a three-phase grid-connected photovoltaic system based on ultra-local model

In this paper, a robust model-free controller for a grid-connected photovoltaic (PV) system is designed. The system consists of a PV generator connected to a three-phase grid by a DC/AC converter. The control objectives of the overall system are to extract maximum power from the PV source, to control reactive power exchange and to improve the quality of the current injected into the grid. The model-free control technique is based on the use of an ultra-local model instead of the dynamic model of the overall system. The local model is continuously updated based on a numerical differentiator using only the input–output behavior of the controlled system. The model-free controller consists of a classical feedback controller and a compensator for the effects of internal parameter changes and external disturbances. Simulation results illustrate the efficiency of the controller for grid-connected PV systems.     

Sputtered Fe1·5CoNi0.5 coating: An improved protective coating for SOFC interconnect applications

In an attempt to optimize the properties of FeCoNi coating for planar solid oxide fuel cell (SOFC) interconnect application, the coating composition is modified by increasing the ratio of Fe/Ni. An Fe_1·5CoNi_0.5 (Fe:Co:Ni = 1.5:1:0.5, atomic ratio) metallic coating is fabricated on SUS 430 stainless steel by magnetron sputtering, followed by oxidation in air at 800°C. The Fe_1·5CoNi_0.5 coating is thermally converted to (Fe,Co,Ni)₃O₄ and (Fe,Co,Mn,Ni)₃O₄ without (Ni,Co)O particles. After oxidation for 1680 h, no further migration of Cr is detected in the thermally converted coating region. A low oxidation rate of 5.9 × 10^?14 g² cm^?4 s^?1 and area specific resistance of 12.64 mΩ·cm² is obtained for Fe_1·5CoNi_0.5 coated steels.     

基于概率法的电力系统安全分析

为应对电力系统安全分析中的停机问题，基于概率法的方式，将常用的确定停机计算与加入了概率法的概率停机进行比较，研究了二者的区别与其在长期投资方向的不同。在进行电力系统停机分析时，通常会分别从确定停机与概率停机的角度出发，对其应急状态下的潮流进行计算。但前者的方法可能导致极低概率的停机事件被忽略，进而影响长期的资金投资。通过加入概率法的计算，使得对单个停机事件的判定由其具体的频率来确定，增加了系统运行的稳定性。     

基于阶段注意力机制的电力线提取算法

电力系统维护是电力系统稳定运行的重要保障，应用智能算法的无人机电力巡检则为电力系统维护提供便捷。电力线提取是自主电力巡检以及保障飞行器低空飞行安全的关键技术，结合深度学习理论进行电力线提取是电力巡检的重要突破点。本文将深度学习方法用于电力线提取任务，结合电力线图像特点嵌入改进的图像输入策略和注意力模块，提出一种基于阶段注意力机制的电力线提取模型（SA-Unet）。本文提出的SA-Unet模型编码阶段采用阶段输入融合策略（Stage input fusion strategy， SIFS），充分利用图像的多尺度信息减少空间位置信息丢失。解码阶段通过嵌入阶段注意力模块（Stage attention module，SAM）聚焦电力线特征，从大量信息中快速筛选出高价值信息。实验结果表明，该方法在复杂背景的多场景中具有良好的性能。     

Morphology-controlled fabrication of nanostructured WO3 thin films by magnetron sputtering with glancing angle deposition for enhanced efficiency photo-electrochemical water splitting

Herein, the tungsten trioxide (WO₃) nanostructure thin films with different morphologies are firstly fabricated by magnetron sputtering with glancing angle deposition technique (MS-GLAD), followed by the post annealed treatment process in air ambient for 2 h. It is demonstrated that the geometry of MS-GLAD setup, mainly substrate position, played a crucial role in determining the morphology, crystallinity, optical transmittance, and photo-electrochemical (PEC) performance of the WO₃ nanostructured thin film. With the different substrate positions in the MS-GLAD system, the WO₃ nanorod film layer could be precisely changed to combine an underlying dense layer with a nanorod layer and then nanocolumnar film. Moreover, the prepared samples' chemical composition and work function are studied by X-ray photoelectron spectroscopy (XPS) and ultraviolet photoelectron spectroscopy (UPS), respectively. The combining WO₃ nanostructure produced high PEC efficiency compared to the single layer of the WO₃ nanorods sample and the dense WO₃ thin film sample. Thus, morphology-controlled nanostructure film based on the MS-GLAD technique in our study provides a simple approach to enhance the photo-anode for PEC water splitting application.