MSC-YOLO:Improved YOLOv7 Based on Multi-Scale Spatial Context for Small Object Detection in UAV-View

Accurately identifying small objects in high-resolution aerial images presents a complex and crucial task in thefield of small object detection on unmanned aerial vehicles(UAVs).This task is challenging due to variations inUAV flight altitude,differences in object scales,as well as factors like flight speed and motion blur.To enhancethe detection efficacy of small targets in drone aerial imagery,we propose an enhanced You Only Look Onceversion 7(YOLOv7)algorithm based on multi-scale spatial context.We build the MSC-YOLO model,whichincorporates an additional prediction head,denoted as P2,to improve adaptability for small objects.We replaceconventional downsampling with a Spatial-to-Depth Convolutional Combination(CSPDC)module to mitigatethe loss of intricate feature details related to small objects.Furthermore,we propose a Spatial Context Pyramidwith Multi-Scale Attention(SCPMA)module,which captures spatial and channel-dependent features of smalltargets acrossmultiple scales.This module enhances the perception of spatial contextual features and the utilizationof multiscale feature information.On the Visdrone2023 and UAVDT datasets,MSC-YOLO achieves remarkableresults,outperforming the baseline method YOLOv7 by 3.0%in terms ofmean average precision(mAP).The MSCYOLOalgorithm proposed in this paper has demonstrated satisfactory performance in detecting small targets inUAV aerial photography,providing strong support for practical applications.

Short Video Recommendation Algorithm Incorporating Temporal Contextual Information and User Context

With the popularity of 5G and the rapid development of mobile terminals,an endless stream of short video software exists.Browsing short-form mobile video in fragmented time has become the mainstream of user’s life.Hence,designing an efficient short video recommendation method has become important for major network platforms to attract users and satisfy their requirements.Nevertheless,the explosive growth of data leads to the low efficiency of the algorithm,which fails to distill users’points of interest on one hand effectively.On the other hand,integrating user preferences and the content of items urgently intensify the requirements for platform recommendation.In this paper,we propose a collaborative filtering algorithm,integrating time context information and user context,which pours attention into expanding and discovering user interest.In the first place,we introduce the temporal context information into the typical collaborative filtering algorithm,and leverage the popularity penalty function to weight the similarity between recommended short videos and the historical short videos.There remains one more point.We also introduce the user situation into the traditional collaborative filtering recommendation algorithm,considering the context information of users in the generation recommendation stage,and weight the recommended short-formvideos of candidates.At last,a diverse approach is used to generate a Top-K recommendation list for users.And through a case study,we illustrate the accuracy and diversity of the proposed method.

How Information Communication Technology(ICT) can Support College Language Teachers in the Bilingual Context

While the increasing development of modern information technology, the globalization is becoming an obvious feature on educational situations. Therefore, mastering some necessary bilingual competencies will present essential meaning for educators. In the case of language teachers who teaching in the ethnically plural countries, for instance, China, the United States, language teachers have to face to various difficulties on the process of teaching in the bilingual class. Currently, the advanced technology is gradually being applied into language teaching, and then provides a series of advantages on improving the quality of language teaching. Firstly, the essay will analyse the barriers which exist in the language class, which in the level of Chinese university. Secondly, it will systematically describe how ICT can help language teachers to solve difficulties on teaching and display diverse innovative technological tools of language teaching.

A New Word Detection Method for Chinese Based on Local Context Information

Finding out out-of-vocabulary words is an urgent and difficult task in Chinese words segmentation. To avoid the defect causing by offline training in the traditional method, the paper proposes an improved prediction by partical match （PPM） segmenting algorithm for Chinese words based on extracting local context information, which adds the context information of the testing text into the local PPM statistical model so as to guide the detection of new words. The algorithm focuses on the process of online segmentatien and new word detection which achieves a good effect in the close or opening test, and outperforms some well-known Chinese segmentation system to a certain extent.

Adopting Context Mediation in Information Integration to Resolve Semantic Heterogeneity in Distributed Environment

Ontology-based semantic information integration resolve the schema-level heterogeneity and part of data level heterogeneity between distributed data sources. But it is ubiquitous that schema semantics of information is identical while the interpretation of it varies with different context, and ontology-based semantic information integration can not resolve this context heterogeneity. By introducing context representation and context mediation to ontology based information integration, the attribute-level context heterogeneity can be detected and reconciled automatically, and hence a complete solution for semantic heterogeneity is formed. Through a concrete example, the context representation and the process in which the attribute-level context heterogeneity is reconciled during query processing are presented. This resolution can make up the deficiency of schema mapping based semantic information integration. With the architecture proposed in this paper the semantic heterogeneity solution is adaptive and extensive.

Mobile and Context-Aware GeoBI Applications: A Multilevel Model for Structuring and Sharing of Contextual Information

With the requirements for high performance results in the today’s mobile, global, highly competitive, and technology-based business world, business professionals have to get supported by convenient mobile decision support systems (DSS). To give an improved support to mobile business professionals, it is necessary to go further than just allowing a simple remote access to a Business Intelligence platform. In this paper, the need for actual context-aware mobile Geospatial Business Intelligence (GeoBI) systems that can help capture, filter, organize and structure the user mobile context is exposed and justified. Furthermore, since capturing, structuring, and modeling mobile contextual information is still a research issue, a wide inventory of existing research work on context and mobile context is provided. Then, step by step, we methodologically identify relevant contextual information to capture for mobility purposes as well as for BI needs, organize them into context-dimensions, and build a hierarchical mobile GeoBI context model which (1) is geo-spatial-extended, (2) fits with human perception of mobility, (3) takes into account the local context interactions and information-sharing with remote contexts, and (4) matches with the usual hierarchical aggregated structure of BI data.

Reasoning about Context Information in Cloud Computing Environments

The notion of context provides flexibility and adaptation to cloud computing services. Location, time identity and activity of users are examples of primary context types. The motivation of this paper is to formalize reasoning about context information in cloud computing environments. To formalize such context-aware reasoning, the logic LCM of context-mixture is introduced based on a Gentzen-type sequent calculus for an extended resource-sensitive logic. LCM has a specific inference rule called the context-mixture rule, which can naturally represent a mechanism for merging formulas with context information. Moreover, LCM has a specific modal operator called the sequence modal operator, which can suitably represent context information. The cut-elimination and embedding theorems for LCM are proved, and a fragment of LCM is shown to be decidable. These theoretical results are intended to provide a logical justification of context-aware cloud computing service models such as a flowable service model.

Fuzzy Privacy Decision for Context-Aware Access Personal Information

A context-aware privacy protection framework was designed for context-aware services and privacy control methods about access personal information in pervasive environment. In the process of user＇s privacy decision, it can produce fuzzy privacy decision as the change of personal information sensitivity and personal information receiver trust. The uncertain privacy decision model was proposed about personal information disclosure based on the change of personal information receiver trust and personal information sensitivity. A fuzzy privacy decision information system was designed according to this model. Personal privacy control policies can be extracted from this information system by using rough set theory. It also solves the problem about learning privacy control policies of personal information disclosure.

An Analysis of the Mode of Information Transfer in High-context Cultures in Terms of Red Cliff

According to Edward Hall,cultures can be categorized as being either high or low context.Chinese culture is typically high-context culture.The paper tries to explore the mode of information transfer in high-context cultures in terms of Red Cliff.The discussion will be focused on dialogues and the costume.This will help understand the characteristics of high-context culture and make the communication between people from cultures of different contexts easier.

论中国语境下Disinformation概念的对接、转换与重新阐释

[目的/意义]针对中西方理解与应用Disinformation概念存在较大差异这一现状,提出中国语境下这一概念的对接、转换与重新阐释,为围绕中国问题开展Disinformation研究创造条件,进而在我国开辟新的研究领域。[方法/过程]梳理分析中西方理解Disinformation概念的差异,然后基于这一概念的3种基本属性,提出Disinformation概念进入中国语境的基本路径,即"对接→转换→重新阐释"。[结果/结论]"信息""误导""意图"3种属性及施众、受众两个要素,是中西方Disinformation概念对接、转换的基础和依据;从"信息"和"信息活动"两个角度,提出中国语境下Disinformation概念的两种定义,Disinformation则译为"误导性信息"或"信息误导与反误导"。Disinformation概念经重新阐释,可应用于研究中国社会现象,解决中国现实问题。

MFFNet:多级特征融合图像语义分割网络

在图像语义分割任务中,大多数方法未将不同尺度、不同层次的特征充分利用就直接进行上采样,会造成一些有效信息被当成冗余信息而被摒弃,从而降低对某些细小类别和相似类别分割的准确性和敏感性。为此,提出一个多级特征融合网络(MFFNet)。MFFNet采用编码器-解码器结构,在编码阶段,通过上下文信息提取路径和空间信息提取路径分别获取上下文信息与空间细节信息,增强像素间关联性与边界准确性;解码阶段设计一条多级特征融合路径,利用混合双边融合模块融合上下文信息;利用高低特征融合模块融合深层信息与空间信息;利用全局通道融合模块获取不同通道之间的联系,实现不同尺度信息的全局融合。MFFNet网络在PASCAL VOC 2012和Cityscapes验证集上的平均交互比(MIoU)分别为80.70%和76.33%,取得了较好的分割结果。

基于上下文信息与特征细化的无人机小目标检测算法

无人机航拍图像中的目标检测是近年来研究的热点,针对无人机视角下目标小而密集、背景复杂导致检测精度低的问题,提出一种基于上下文信息与特征细化的无人机小目标检测算法。通过上下文特征增强模块,利用多尺度扩张卷积捕获与周围区域像素点的潜在关系,为网络补充上下文信息,并根据不同尺度的特征层自适应生成各层级特征图的输出权重,动态优化特征图表达能力;由于不同特征图细粒度不同,使用特征细化模块来抑制特征融合中冲突信息的干扰,防止小目标特征淹没在冲突信息中;设计了一种带权重的损失函数,加快模型收敛速度,进一步提高小目标检测精度。在VisDrone2021数据集进行大量实验表明,改进后的模型较基准模型mAP50提高8.4个百分点,mAP50:95提高5.9个百分点,FPS为42,有效提高了无人机航拍图像中小目标的检测精度。

融合实体和上下文信息的篇章关系抽取研究

篇章关系抽取旨在识别篇章中实体对之间的关系.相较于传统的句子级别关系抽取,篇章级别关系抽取任务更加贴近实际应用,但是它对实体对的跨句子推理和上下文信息感知等问题提出了新的挑战.本文提出融合实体和上下文信息(Fuse entity and context information,FECI)的篇章关系抽取方法,它包含两个模块,分别是实体信息抽取模块和上下文信息抽取模块.实体信息抽取模块从两个实体中自动地抽取出能够表示实体对关系的特征.上下文信息抽取模块根据实体对的提及位置信息,从篇章中抽取不同的上下文关系特征.本文在三个篇章级别的关系抽取数据集上进行实验,效果得到显著提升.

结合混合卷积和多尺度注意力的视频异常检测算法

基于U-net风格的无监督视频异常检测模型有着较好的检测效果,但由于普通卷积运算使用固有的局部特性,使U-Net风格的编码器无法有效地提取全局上下文信息,并且使用简单的跳跃连接无法获得有效的特征信息,使用的L2损失函数是仅考虑了像素级别的差异而无法捕捉图像的结构特征。对此提出了结合混合卷积和多尺度注意力的视频异常检测算法,并加入结构相似性损失函数(SSIM)优化模型。具体来说,在编码器最后一层添加混合卷积模块,混合空间和位置的特征来提取全局上下文信息。在编码器和解码器之间的跳跃连接中添加多尺度注意力模块,使模型能提取更有价值的特征,实现有效的跳跃连接。使用参数约束结构相似性损失函数与L2损失函数的权重,从而更准确地优化模型。实验结果表明,所提算法在UCSD-Ped2和CUHK Avenue公开数据集上的AUC指标达到96.7%和86.1%,与改进前的模型相比提高了1.6%和1.4%,证明了所提模型的有效性。

改进Faster R-CNN的遥感图像小目标检测算法

遥感图像目标检测是目标检测领域的一个关键问题,目前利用深度学习检测目标的算法大多在单向特征融合过程中添加注意力机制,一视同仁地去增强各类型的目标,并不能突出小目标。为了取得更好的检测效果,通过引入非对称高低层调制机制,构造兼顾低层细节信息和高层语义信息的特征图,以达到增强小目标特征检测的目的;同时使用DIoU损失函数代替原算法SmoothL1损失函数以提升算法检测精度与收敛速度;并且在感兴趣区域分类任务中引入灵活上下文信息以提高小目标分类准确性。实验结果表明,该算法在DIOR和NWPU VHR-10数据集上均取得了良好的表现。

CIEFRNet:面向高速公路的抛洒物检测算法

高速公路抛洒物危及行车安全,极易诱发交通事故,及时识别并清理高速公路抛洒物十分重要。由于高速公路抛洒物在图像中面积占比小且图像背景复杂,现有检测方法常出现漏检和误检的情况。针对上述问题,提出了一种基于上下文信息增强和特征提纯的抛洒物检测算法,记为CIEFRNet。设计了一种融合上下文Transformer的主干特征提取模块(CSP-COT),充分挖掘局部静态上下文信息和全局动态上下文信息,增强小抛洒物的特征表示;主干网络中使用改进的空间金字塔池化(ISPP),通过级联的空洞卷积实现特征的多尺度下采样,减轻目标细节信息的损失;为提高特征融合能力,设计了特征提纯模块(CNAB),其中嵌入了提出的一种混合注意力机制(ECSA),可抑制图像背景噪声,强化微小抛洒物的特征;引入基于动态非单调聚焦机制的WIoU优化损失函数,提高小抛洒物学习能力,加速网络收敛。实验结果表明,所提方法在自制的高速公路抛洒物数据集上的精确率、召回率、AP0.5和AP0.5:0.95分别达到96.5%、81.6%、88.1%和46.5%,优于当前主流的目标检测方法,其算法复杂度也更低,满足实际场景应用需要。

基于分层融合策略和上下文信息嵌入的多模态情绪识别

现有的多模态融合策略大多将不同模态特征进行简单拼接,忽略了针对单个模态固有特点的个性化融合需求。同时,在情绪识别阶段,独立地看待单个话语的情绪而不考虑其在前后话语语境下的情绪状态,可能导致情绪识别错误。为了解决上述问题,提出一种基于分层融合策略和上下文信息嵌入的多模态情绪识别方法,通过分层融合策略,采用层次递进的方式,依次融合不同的模态特征,以便减少单个模态的噪声干扰并解决不同模态间表达不一致的问题。该方法还充分利用融合后模态的上下文信息,综合考虑单个话语在上下文语境中的情绪表示,以便提升情绪识别的效果。在二分类情绪识别任务中,该方法的准确率比SOTA模型提升1.54%。在多分类情绪识别任务中,该方法的F1值比SOTA模型提升2.79%。

海量层次信息的Focus+Context交互式可视化技术

综述了海量层次信息可视化与Focus+Context技术的相关工作,针对海量层次信息可视化的交互问题,在嵌套圆可视化技术的基础上提出了基于上下文感知的Focus+Context交互式可视化技术.首先,基于外切圆排列方法提出对圆心进行三角网格剖分的方法,为变形计算建立上下文;然后,针对变形计算前后上下文一致性问题,在三角网格邻居跟踪方法的基础上,提出了用于同层兄弟节点上下文感知的外切圆变形排列方法,以及用于父子节点上下文感知的嵌套圆迭代排列方法.实验结果表明。上述方法在实现焦点突出的鱼眼视图的同时,能够有效地解决Focus+Context交互式可视化的上下文感知问题.上述方法应用于文件系统海量层次信息的交互式可视化问题,提供了交互式可视化工具.

面向信息可视化的语义Focus+Context人机交互技术

大数据成为继云计算和物联网之后,国际学术界和产业界所共同瞩目的又一个研究热点.信息可视化是辅助用户洞悉大数据背后隐藏的知识和规律的重要方法和有力工具.如何在图形用户界面中对大规模信息以符合认知规律的方式进行可视化,并且使得计算机能够智能化的理解用户意图以配合其进行高效的人机交互,是信息可视化面临的挑战之一.文中提出一种面向信息可视化的语义Focus+Context人机交互技术.首先,在基于空间距离的经典Focus+Context数学模型基础上对其进行语义建模和扩展,建立了面向信息空间和可视化表征空间的语义距离模型以及语义关注度模型,定义了交互中的焦点对象与语义上下文.其次,在此基础上建立了语义Focus+Context用户界面模型,给出了界面抽象元素和实体元素以及映射关系的形式化描述,同时建立了Focus+Context交互循环机制.最后,给出了应用于经典Focus+Context及鱼眼数学模型的描述,表明文中提出方法具有很好的兼容性描述能力;同时,给出了面向文件系统主题聚集的语义Focus+Context应用,给出了基于主题语义关注度与嵌套圆鱼眼视图的动态可视化实例,应用实例表明文中提出技术能够有效支持用户在信息可视化界面中对大规模信息进行智能化的可视化和交互探索.

集成多种上下文与混合交互的显著性目标检测

显著性目标检测目的是识别和分割图像中的视觉显著性目标,它是计算机视觉任务及其相关领域的重要研究内容之一。当下基于全卷积网络(FCNs)的显著性目标检测方法已经取得了不错的性能,然而现实场景中的显著性目标类型多变且尺寸不固定,这使得准确检测并完整分割出显著性目标仍然是一个巨大的挑战。为此,该文提出集成多种上下文和混合交互的显著性目标检测方法,通过利用密集上下文信息探索模块和多源特征混合交互模块来高效预测显著性目标。密集上下文信息探索模块采用空洞卷积、不对称卷积和密集引导连接渐进地捕获具有强关联性的多尺度和多感受野上下文信息,通过集成这些信息来增强每个初始多层级特征的表达能力。多源特征混合交互模块包含多种特征聚合操作,可以自适应交互来自多层级特征中的互补性信息,以生成用于准确预测显著性图的高质量特征表示。此方法在5个公共数据集上进行了性能测试,实验结果表明,该文方法在不同的评估指标下与19种基于深度学习的显著性目标检测方法相比取得优越的预测性能。