基于改进MobileFaceNet的人脸识别方法北大核心

针对目前大多数人脸识别算法参数多、计算量大,难以部署到移动端和嵌入式设备中的问题,提出了一种基于改进MobileFaceNet的人脸识别方法。通过对MobileFaceNet模型结构的调整,将bottleneck模块优化为sandglass模块,改良深度卷积和逐点卷积的相对位置,适当增大sandglass模块的输出通道数,从而减少特征压缩时的信息丢失,增强人脸空间特征的提取。实验结果表明:改进后的方法在LFW测试数据集上准确率达99.15%,模型大小和计算量分别仅为原算法的61%和45%,验证了所提方法的有效性。     

从蚕的生命到艺术：中国古代生命观的哲学表达

中国考古出土的蚕业实物及蚕的艺术形象比较丰富,蚕的艺术形象如蚕纹、陶蚕蛹、牙雕蚕、玉石蚕、铜蚕、金蚕等,可统称为"蚕的模拟形态"。对蚕的模拟形态的功用,已有的诸多解释都有待完善。研究表明,蚕的模拟形态或艺术形象表达的功用或为饰品,或为装饰图案,或有待进一步考究。但无论哪种功用,用"蚕"这一形象都蕴含了特有的用意。通过对中国古代生命观的考察,文章认为蚕的艺术形象折射出相应的中国古代哲学生命观,即中国古人追求的死而复生、生生不息、羽化成仙、长乐无极等观念。     

高面板坝全寿命周期变形控制方法及应用

高面板坝的变形对面板的安全运行有着特别重要的影响，国内外已建的高面板坝工程中，因坝体变形大导致防渗面板挤压破损，坝体渗漏量大的实例较多，不得不降低水库水位进行修复处理，造成较大的经济损失乃至给大坝的长期运行留下安全隐患。通过发生挤压破损的实例分析，发现变形控制缺乏系统性是发生面板挤压破损的主要因素，为预防面板破损，系统提出了“控制坝体总变形，转化有害变形，适应纵向变形”的坝体变形控制方法，并在使用软硬岩混合料筑坝的董箐面板堆石坝中得到的应用，取得了良好效果，该工程运行至今达十余年，未见面板有挤压破损迹象，该方法对建设200 m以上乃至300 m级超高面板坝具有重要借鉴意义。     

Differential gene expression of the healthy conjunctiva during the day

PurposeTo determine if there is diurnal variation in gene expression in normal healthy conjunctival cells.MethodsBulbar conjunctival swab samples were collected from four healthy subjects in the morning and evening of the same day. The two swab samples were taken from one eye of each participant, with a minimum of five hours gap between the two samples. RNA was extracted and analysed using RNA sequencing (RNA-Seq).ResultsA total of 121 genes were differentially expressed between the morning and the evening conjunctival samples, of which 94 genes were upregulated in the morning, and 27 genes were upregulated in the evening. Many of the genes that were upregulated in the morning were involved in defence, cell turnover and regulation of gene expression, while the genes upregulated in the evening were involved in signalling and mucin production.ConclusionsThis study has identified several genes whose expression changes over the course of the day. Knowledge of diurnal variations of conjunctival gene expression provides an insight into the regulatory status of the healthy eye and provides a baseline for examining changes during ocular surface disease.     

A Smart Deep Convolutional Neural Network for Real-Time Surface Inspection

A proper detection and classification of defects in steel sheets in real time have become a requirement for manufacturing these products, largely used in many industrial sectors. However, computers used in the production line of small to medium size companies, in general, lack performance to attend real-time inspection with high processing demands. In this paper, a smart deep convolutional neural network for using in real-time surface inspection of steel rolling sheets is proposed. The architecture is based on the state-of-the-art SqueezeNet approach, which was originally developed for usage with autonomous vehicles. The main features of the proposed model are: small size and low computational burden. The model is 10 to 20 times smaller when compared to other networks designed for the same task, and more than 700 times smaller than general networks. Also, the number of floating-point operations for a prediction is about 50 times lower than the ones used for similar tasks. Despite its small size, the proposed model achieved near-perfect accuracy on a public dataset of 1800 images of six types of steel rolling defects.     

In Vivo Simultaneous Analysis of Gene Expression by Dual-Color Luciferases in Caenorhabditis elegans

Both fluorescent and luminescent observation are widely used to examine real-time gene expression patterns in living organisms. Several fluuorescent and luminescent proteins with specific optical properties have been developed and applied for simultaneous, multi-color observation of more than two gene expression profiles. Compared to fluorescent proteins, however, the application of multi-color luminescent imaging in living organisms is still limited. In this study, we introduced two-color luciferases into the soil nematode C. elegans and performed simultaneous analysis of two gene expression profiles. Using a green-emitting luciferase Eluc (emerald luciferase) and red-emitting luciferase SLR (stable luciferase red), the expression patterns of two genes were simultaneously observed in single animals from embryonic to adult stages over its whole life span. In addition, dual gene activities were observed at the single embryo level, with the simultaneous observation of morphological changes. These are the first application of a two-color luciferase system into a whole animal and suggest that precise relationship of expression patterns of multiple genes of interest can be analyzed over the whole life of the animal, dependent on the changes in genetic and/or environmental conditions.     

Sequential alignment attention model for scene text recognition

Scene text recognition has been a hot research topic in computer vision due to its various applications. The state-of-the-art solutions usually depend on the attention-based encoder-decoder framework that learns the mapping between input images and output sequences in a purely data-driven way. Unfortunately, there often exists severe misalignment between feature areas and text labels in real-world scenarios. To address this problem, this paper proposes a sequential alignment attention model to enhance the alignment between input images and output character sequences. In this model, an attention gated recurrent unit (AGRU) is first devised to distinguish the text and background regions, and further extract the localized features focusing on sequential text regions. Furthermore, CTC guided decoding strategy is integrated into the popular attention-based decoder, which not only helps to boost the convergence of the training but also enhances the well-aligned sequence recognition. Extensive experiments on various benchmarks, including the IIIT5k, SVT, and ICDAR datasets, show that our method substantially outperforms the state-of-the-art methods.