Using attention LSGB network for facial expression recognition

Pattern Analysis and Applications - Both the multiple sources of the available in-the-wild datasets and noisy information of images lead to huge challenges for discriminating subtle distinctions...     

IoT设备程序同源性智能检测技术综述

IoT设备与各行业的深度融合方兴日盛,使得IoT程序快速开发的需求不断增长。开发者习惯于集成第三方库或常用代码。不幸的是,若这些代码中隐藏着潜在的漏洞,那他们也会被扩散到不同的程序中,为其大规模扩散创造了条件。这也是造成近年来IoT设备群体性安全事件频发的重要原因之一。为了降低危害,发现具备相似漏洞的程序并进行相关的处置是一个有效的方法。同源性分析作为挖掘程序间关联关系的重要手段之一,可高效地实现程序漏洞的智能溯源取证。结合机器学习和深度学习技术,它表现出解决大规模程序安全性检测的巨大潜力。然而,IoT设备的软硬件特点仍使得该技术的使用面临挑战。当前已有诸多方案在IoT设备程序的同源性智能检测方面取得了进展。因此,本文将系统性回顾近年来相关技术研究的成果,将他们分为相似性分析和创作者归属技术。首先,我们介绍了两种方式的数据来源。接着,检测过程中涉及的特征选择、特征表示以及相对应的检测方法也被依次介绍。进一步的,本文不仅比较和总结了方案的特点和局限性,还对他们在不同类型IoT设备程序的适配性进行了对比分析。最后,文章针对IoT程序分析提出了一些研究建议。作者希望本综述可为研究者阐明这些工作的核心技术点,并为他们在IoT设备上的进一步应用提供启发。     

大型光学遥感卫星高可靠高性能综合电子系统设计

针对大型光学遥感卫星高可靠高性能和好用易用的需求,设计了一种具有高度容错能力的综合电子系统,详细介绍了 系统的体系架构和信息流设计,深入探讨了卫星运行管理和系统容错机制等关键设计技术。 介绍了所设计综合电子系统在某 型号卫星上的在轨应用情况,该卫星充分采用了论文提出的多冗余备份架构、容错技术和运行管理技术,平均每天完成 21 次成 像和 11 次数传,日均拍摄面积高达 1 100 000 km 2 ,最长连续无故障运行时间超过 6 个月,充分验证了所设计综合电子系统具有 在轨高效稳定运行的能力。     

Face anti-spoofing based on multi-modal and multi-scale features fusion

Face anti-spoofing is used to assist face recognition system to judge whether the detected face is real face or fake face. In the traditional face anti-spoofing methods, features extracted by hand are used to describe the difference between living face and fraudulent face. But these handmade features do not apply to different variations in an unconstrained environment. The convolutional neural network (CNN) for face deceptions achieves considerable results. However, most existing neural network-based methods simply use neural networks to extract single-scale features from single-modal data, while ignoring multi-scale and multi-modal information. To address this problem, a novel face anti-spoofing method based on multi-modal and multi-scale features fusion ( MMFF) is proposed. Specifically, first residual network ( Resnet )-34 is adopted to extract features of different scales from each modality, then these features of different scales are fused by feature pyramid network (FPN), finally squeeze-and-excitation fusion ( SEF) module and self-attention network ( SAN) are combined to fuse features from different modalities for classification. Experiments on the CASIA-SURF dataset show that the new method based on MMFF achieves better performance compared with most existing methods.     

基于相对生成对抗网络的低清小脸幻构

针对当前代表性低清小脸幻构方法存在的视觉真实感弱、网络结构复杂等问题,提出了一种基于相对生成对抗网络的低清小脸幻构方法(tfh-RGAN).该文方法的网络架构包括幻构生成器和判别器两个部分,通过像素损失函数和相对生成对抗损失函数的联合最小化,实现生成器和判别器的交替迭代训练.其中,幻构生成器结合了残差块、稠密块以及深度可分离卷积算子,保证幻构效果和网络深度的同时降低生成器的参数量;判别器采用图像分类问题中的全卷积网络,通过先后去除批归一化层、添加全连接层,充分挖掘相对生成对抗网络在低清小脸幻构问题上的能力极限.实验结果表明,在不额外显式引入任何人脸结构先验的条件下,该文方法能够以更简练的网络架构输出清晰度更高、真实感更强的幻构人脸.从定量角度看,该文方法的峰值信噪比相较之前的若干代表性方法可提高0.25～1.51 dB.     

Insights into the Influence of Work Functions of Cathodes on Efficiencies of Perovskite Solar Cells

Though various efforts on modification of electrodes are still undertaken to improve the efficiency of perovskite solar cells, attributing to the large scope of these methods, it is of significance to unveil the working principle systematically. Herein, inverted perovskite solar cells based on indium tin oxide (ITO)/poly(3,4‐ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS)/CH₃NH₃PbI₃/phenyl‐C61‐butyric acid methyl ester (PC₆₁ BM)/buffer metal/Al are constructed. Through the choice of different buffer metals to tune work function of the cathode, the contact nature of the active layer with the cathode could be manipulated well. In comparison with the device using Au/Al as the electrode that shows an unfavorable band bending for conducting the excited electrons to the cathode, the one with Ca/Al presents a dramatically improved efficiency over 17.1%, ascribed to the favorable band bending at the interface of the cathode with the active layer. Details for tuning the band bending and the corresponding charge transfer mechanism are given in a systematic manner. Thus, a general guideline for constructing perovskite photovoltaic devices efficiently is provided.     

Polypyrrole Whelk‐Like Arrays toward Robust Controlling Manipulation of Organic Droplets Underwater

Whelk‐like polypyrrole (PPy) arrays film is successfully prepared by electropolymerization of pyrrole in the presence of low‐surface‐energy tetraethylammonium perfluorooctanesulfonate (TEAPFOS) as dopant. The underwater wettability of PPy whelk‐like arrays can be successfully tuned by electrical doping/dedoping of PFOS ions. Interestingly, CCl₄ droplets with microliter‐size as a representative sample are gathered together to form a larger droplet underwater at the potential of +0.8 V (vs Ag/AgCl), because PPy is in its PFOS‐doped states. Note that CCl₄ droplet can climb uphill successfully on the inclined whelk‐like arrays PPy film under the applied potential of ?1.0 V (vs Ag/AgCl), which may be attributed to wettability gradient derived from different oxidation states of PPy induced by electrochemical potential. These results may provide a simple strategy for on‐demand manipulation of organic droplets underwater at low voltage.     

Epitaxial growth of large-area and highly crystalline anisotropic ReSe<Subscript>2</Subscript> atomic layer

The anisotropic two-dimensional (2D) layered material rhenium disulfide (ReSe2) has attracted considerable attention because of its unusual properties and promising applications in electronic and optoelectronic devices.However,because of its low lattice symmetry and interlayer decoupling,anisotropic growth and out-of-plane growth occur easily,yielding thick flakes,dendritic structure,or flower-like structure.In this study,we demonstrated a bottom-up method for the controlled and scalable synthesis of ReSe2 by van der Waals epitaxy.To achieve controllable growth,a micro-reactor with a confined reaction space was constructed by stacking two mica substrates in the chemical vapor deposition system.Within the confined reaction space,the nucleation density and growth rate of ReSe2 were significantly reduced,favoring the large-area synthesis of ReSe2 with a uniform monolayer thickness.The morphological evolution of ReSe2 with growth temperature indicated that the anisotropic growth was suppressed at a low growth temperature (＜600 ℃).Field-effect transistors employing the grown ReSe2 exhibited p-type conduction with a current ON/OFF ratio up to 10s and a hole carrier mobility of 0.98 cm2/(V.s).Furthermore,the ReSe2 device exhibited an outstanding photoresponse to near-infrared light,with responsivity up to 8.4 and 5.1 A/W for 850-and 940-nm light,respectively.This work not only promotes the large-scale application of ReSe2 in high-performance electronic devices but also clarifies the growth mechanism of low-lattice symmetry 2D materials.