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.     

使能未来工厂的5G能力综述

5G系统将移动通信服务从移动电话、移动宽带和大规模机器通信扩展到新的应用领域，即所谓对通信服务有特殊要求的垂直领域。对使能未来工厂的5G能力进行了全面的分析总结，包括弹性网络架构、灵活频谱、超可靠低时延通信、时间敏感网络、安全和定位，而弹性网络架构又包括对网络切片、非公共网络、5G局域网和边缘计算的支持。希望从广度到深度，对相关的理论及技术应用做透彻、全面的梳理，对其挑战做清晰的总结，从而为相关研究和工程技术人员提供借鉴。     

Understanding the asymmetric perceptions of smartphone security from security feature perspective: A comparative study

Smartphones are being used and relied on by people more than ever before. The open connectivity brings with it great convenience and leads to a variety of risks that cannot be overlooked. Smartphone vendors, security policy designers, and security application providers have put a variety of practical efforts to secure smartphones, and researchers have conducted extensive research on threat sources, security techniques, and user security behaviors. Regrettably, smartphone users do not pay enough attention to mobile security, making many efforts futile. This study identifies this gap between technology affordance and user requirements, and attempts to investigate the asymmetric perceptions toward security features between developers and users, between users and users, as well as between different security features. These asymmetric perceptions include perceptions of quality, perceptions of importance, and perceptions of satisfaction. After scoping the range of smartphone security features, this study conducts an improved Kano-based method and exhaustively analyzes the 245 collected samples using correspondence analysis and importance satisfaction analysis. The 14 security features of the smartphone are divided into four Kano quality types and the perceived quality differences between developers and users are compared. Correspondence analysis is utilized to capture the relationship between the perceived importance of security features across different groups of respondents, and results of importance-satisfaction analysis provide the basis for the developmental path and resource reallocation strategy of security features. This article offers new insights for researchers as well as practitioners of smartphone security.     

Achieving Rich and Active Alkaline Hydrogen Evolution Heterostructures via Interface Engineering on 2D 1T‐MoS2 Quantum Sheets

Large‐scale production of hydrogen from water‐alkali electrolyzers is impeded by the sluggish kinetics of hydrogen evolution reaction (HER) electrocatalysts. The hybridization of an acid‐active HER catalyst with a cocatalyst at the nanoscale helps boost HER kinetics in alkaline media. Here, it is demonstrated that 1T–MoS₂ nanosheet edges (instead of basal planes) decorated by metal hydroxides form highly active ${\text{edge}}_{{\text{1T‐MoS}}_{\text{2}}}$/ ${\text{edge}}_{\text{Ni}{(\text{OH})}_{\text{2}}}$ heterostructures, which significantly enhance HER performance in alkaline media. Featured with rich ${\text{edge}}_{{\text{1T‐MoS}}_{\text{2}}}$/ ${\text{edge}}_{\text{Ni}{(\text{OH})}_{\text{2}}}$ sites, the fabricated 1T–MoS₂ QS/Ni(OH)₂ hybrid (quantum sized 1T–MoS₂ sheets decorated with Ni(OH)₂ via interface engineering) only requires overpotentials of 57 and 112 mV to drive HER current densities of 10 and 100 mA cm^?2, respectively, and has a low Tafel slope of 30 mV dec^?1 in 1 m KOH. So far, this is the best performance for MoS₂‐based electrocatalysts and the 1T–MoS₂ QS/Ni(OH)₂ hybrid is among the best‐performing non‐Pt alkaline HER electrocatalysts known. The HER process is durable for 100 h at current densities up to 500 mA cm^?2. This work not only provides an active, cost‐effective, and robust alkaline HER electrocatalyst, but also demonstrates a design strategy for preparing high‐performance catalysts based on edge‐rich 2D quantum sheets for other catalytic reactions.     

基于深度学习的秀丽隐杆线虫显微图像分割方法

利用计算机实现自动、准确的秀丽隐杆线虫(C.elegans)的各项形态学参数分析,至关重要的是从显微图像上分割出线虫体态,但由于显微镜下的图像噪声较多,线虫边缘像素与周围环境相似,而且线虫的体态具有鞭毛和其他附着物需要分离,多方面因素导致设计一个鲁棒性的C.elegans分割算法仍然面临着挑战。针对这些问题,提出了一种基于深度学习的线虫分割方法,通过训练掩模区域卷积神经网络(Mask R-CNN)学习线虫形态特征实现自动分割。首先,通过改进多级特征池化将高级语义特征与低级边缘特征融合,结合大幅度软最大损失(LMSL)损失算法改进损失计算;然后,改进非极大值抑制;最后,引入全连接融合分支等方法对分割结果进行进一步优化。实验结果表明,相比原始的Mask R-CNN,该方法平均精确率(AP)提升了4.3个百分点,平均交并比(mIOU)提升了4个百分点。表明所提出的深度学习分割方法能够有效提高分割准确率,在显微图像中更加精确地分割出线虫体。     

Rectangular Section Control Technology for Silicon Steel Rolling

Rectangular section control technology(RSCT)was introduced to achieve high-precision profile control during silicon steel rolling.The RSCT principle and method were designed,and the whole RSCT control strategy was developed.Specifically,RSCT included roll contour design,rolling technology optimization,and control strategy development,aiming at both hot strip mills(HSMs)and cold strip mills(CSMs).Firstly,through the high-performance variable crown(HVC)work roll optimization design in the upper-stream stands and the limited shifting technology for schedule-free rolling in the downstream stands of HSMs,a hot strip with a stable crown and limited wedge,local spot,and single wave was obtained,which was suitable for cold rolling.Secondly,an approximately rectangular section was obtained by edge varying contact(EVC)work roll contour design,edge-drop setting control,and closed loop control in the upper-stream stands of CSMs.Moreover,complex-mode flatness control was realized by coordinating multiple shape-control methods in the downstream stands of CSMs.In addition,the RSCT approach was applied in several silicon-steel production plants,where an outstanding performance and remarkable economic benefits were observed.     

移动边缘计算环境下的动态资源分配策略

在通讯设备爆炸式增长的时代,移动边缘计算作为5G通讯技术的核心技术之一,对其进行合理的资源分配显得尤为重要。移动边缘计算的思想是把云计算中心下沉到基站部署(边缘云),使云计算中心更加靠近用户,以快速解决计算资源分配问题。但是,相对于大型的云计算中心,边缘云的计算资源有限,传统的虚拟机分配方式不足以灵活应对边缘云的计算资源分配问题。为解决此问题,提出一种根据用户综合需求变化的动态计算资源和频谱分配算法(DRFAA),采用"分治"策略,并将资源模拟成"流体"资源进行分配,以寻求较大的吞吐量和较低的传输时延。实验仿真结果显示,动态计算资源和频谱分配算法可以有效地降低用户与边缘云之间的传输时延,也可以提高边缘云的吞吐量。     

Specific emitter identification for satellite communication using probabilistic neural networks

Electromagnetic signal emitted by satellite communication (satcom) transmitters are used to identify specific individual uplink satcom terminals sharing the common transponder in real environment, which is known as specific emitter identification (SEI) that allows for early indications and warning (I&W) of the targets carrying satcom furnishment and furthermore the real time electromagnetic situation awareness in military operations. In this paper, the authors are the first to propose the identification of specific transmitters of satcom by using probabilistic neural networks (PNN) to reach the goal of target recognition. We have been devoted to the examination by exploring the feasibility of utilizing the Hilbert transform to signal preprocessing, applying the discrete wavelet transform to feature extraction, and employing the PNN to perform the classification of stationary signals. There are a total of 1000 sampling time series with binary phase shift keying (BPSK) modulation originated by five types of satcom transmitters in the test. The established PNNs classifier implements the data testing and finally yields satisfactory accuracy at 8 dB(±1 dB) carrier to noise ratio, which indicates the feasibility of our method, and even the keen insight of its application in military.     

Clinical significance of contact lens related changes of ocular surface tissue observed on optical coherence images

PurposeTo investigate the relationship between the real contact lens imprint into the conjunctival tissue, observed by optical coherence tomography (OCT) and conjunctival staining and contact lens wearing comfort.Methods17 participants (mean age = 26.6 SD ± 3.6 years; 7 females) were fitted with three different contact lenses base curves of the same silicone hydrogel custom lens type (Visell 50; Hecht Contactlinsen, Au, Germany) in a randomised order. One lens was optimally fitted according to the manufacturer's recommendation, one fitted 0.4 mm flatter and one fitted 0.4 mm steeper. After 4 h of lens wear the contact lens edge in the area of the conjunctiva was imaged nasally and temporally using OCT (Optovue iVue SD-OCT). To correct the artefact due to optical distortion with OCT, the imprint of all worn lenses was measured on a glass plate afterwards. Conjunctival staining in the limbal region after 4 h of lens wear was classified using the CCLRU Grading Scale. Comfort scoring was based on visual analog scales from 0 (very poor) to 100 (excellent).ResultsThe mean conjunctival imprint of all contact lens edges was 32.0 ± 8.1 μm before and 7.3 ± 6.5 μm after distortion correction of the OCT images. The distortion corrected conjunctival imprint with the 0.4 mm steeper lens (11.5 ± 6.2 μm) was statistically significantly greater compared to the optimally fitted lens (6.5 ± 5.9 μm) (One-way ANOVA followed Tukey-test; p = 0.017) and greater compared to the 0.4 mm flatter lens (3.9 ± 5.3 μm) (p < 0.001). There was no statistically significant difference between the optimally fitted lens and the 0.4 mm flatter lens (p = 0.209). The nasally measured imprint (11.4 ± 9.0 μm) was significantly greater than the temporally measured (3.3 ± 7.6 μm) (p < 0.001). There was no statistically significant correlation between the amount of conjunctival imprint and the graded conjunctival staining (p = 0.346) or the wearer’s comfort (p = 0.735).ConclusionsContact lens edges imaged by OCT exhibited displacement artefacts. The observed conjunctival imprints are a combination of real conjunctival compression and artefacts. A deeper imprint of the contact lens into the conjunctiva caused by a steeper base curve was not related to clinically significant staining or changes in comfort after 4 h of lens wear. The observed differences between nasal and temporal imprint are likely to be caused by variations of conjunctival thickness and the shape of the underlying sclera.     

A hybrid multiple feature construction approach for classification using Genetic Programming

The purpose of feature construction is to create new higher-level features from original ones. Genetic Programming (GP) was usually employed to perform feature construction tasks due to its flexible representation. Filter-based approach and wrapper-based approach are two commonly used feature construction approaches according to their different evaluation functions. In this paper, we propose a hybrid feature construction approach using genetic programming (Hybrid-GPFC) that combines filter’s fitness function and wrapper’s fitness function, and propose a multiple feature construction method that stores top excellent individuals during a single GP run. Experiments on ten datasets show that our proposed multiple feature construction method (Fcm) can achieve better (or equivalent) classification performance than the single feature construction method (Fcs), and our Hybrid-GPFC can obtain better classification performance than filter-based feature construction approaches (Filter-GPFC) and wrapper-based feature construction approaches (Wrapper-GPFC) in most cases. Further investigations on combinations of constructed features and original features show that constructed features augmented with original features do not improve the classification performance comparing with constructed features only. The comparisons with three state-of-art methods show that in majority of cases, our proposed hybrid multiple feature construction approach can achieve better classification performance.