针对带钢表面缺陷在实际场景中检测精度低,易出现漏检和误检的情况,构建一种YOLOv5-CFD模型对带钢缺陷目标进行更精确的检测,该模型由CSPDarknet53、FcaNet与解耦检测头(Decoupled head)组成。首先,采用模糊C均值(FCM)算法对东北大学公...针对带钢表面缺陷在实际场景中检测精度低,易出现漏检和误检的情况,构建一种YOLOv5-CFD模型对带钢缺陷目标进行更精确的检测,该模型由CSPDarknet53、FcaNet与解耦检测头(Decoupled head)组成。首先,采用模糊C均值(FCM)算法对东北大学公开的NEU-DET热轧带钢表面缺陷检测数据集中的锚框进行聚类,优化先验框和真实框之间的匹配度;其次,为提取目标区域丰富的细节信息,在原始YOLOv5算法基础上添加频域通道注意力模块FcaNet;最后,采用解耦检测头将分类任务和回归任务分离。在NEU-DET数据集上的实验结果表明,改进的YOLOv5算法在引入较少参数量的情况下,检测精度提高了4.2个百分点,平均精度均值(mAP)达到85.5%,每秒传输帧数(Frames Per Second,FPS)达到27.71,与原YOLOv5相差不大,能够满足检测实时性的要求。展开更多
Nowadays, crop diseases are a crucial problem to the world’s food supplies, in a world where the population count is around 7 billion people, with more than 90% not getting access to the use of tools or features that...Nowadays, crop diseases are a crucial problem to the world’s food supplies, in a world where the population count is around 7 billion people, with more than 90% not getting access to the use of tools or features that would identify and solve the problem. At present, we live in a world dominated by technology on a significant scale, major network coverage, high-end smart-phones, as long as there are great discoveries and improvements in AI. The combination of high-end smart-phones and computer vision via Deep Learning has made possible what can be defined as “smartphone-assisted disease diagnosis”. In the area of Deep Learning, multiple architecture models have been trained, some achieving performance reaching more than 99.53% [1]. In this study, we evaluate CNN’s architectures applying transfer learning and deep feature extraction. All the features obtained will also be classified by SVM and KNN. Our work is feasible by the use of the open source Plant Village Dataset. The result obtained shows that SVM is the best classifier for leaf’s diseases detection.展开更多
考虑到无人机群在协同完成任务时对时延的高要求,选用先验式路由协议OLSR(Optimized Link State Routing)协议。但无人机自组网中无人机节点高速移动和能量有限的特性,使得OLSR选举出来的MPR(Multi-Point Relay)节点可能会因此而丧失MP...考虑到无人机群在协同完成任务时对时延的高要求,选用先验式路由协议OLSR(Optimized Link State Routing)协议。但无人机自组网中无人机节点高速移动和能量有限的特性,使得OLSR选举出来的MPR(Multi-Point Relay)节点可能会因此而丧失MPR资格,从而导致时延增加,网络开销增大。针对该问题,提出一种基于节点速度和能量的MPR集选择算法,运用HELLO分组在邻居探测的过程中感知节点能量和速度,之后在MPR选举前根据节点速度和能量对一跳邻居进行预处理,从而使速度快能量低的节点永不成为MPR节点。排除掉节点后,在节点意愿值相同的情况下再次对节点的速度和能量进行加权计算,选出最优MPR节点。仿真结果表明,基于节点速度和能量的MPR集选择算法在时延、吞吐量、节点能量消耗三个指标都具有良好的特性。展开更多
文摘针对带钢表面缺陷在实际场景中检测精度低,易出现漏检和误检的情况,构建一种YOLOv5-CFD模型对带钢缺陷目标进行更精确的检测,该模型由CSPDarknet53、FcaNet与解耦检测头(Decoupled head)组成。首先,采用模糊C均值(FCM)算法对东北大学公开的NEU-DET热轧带钢表面缺陷检测数据集中的锚框进行聚类,优化先验框和真实框之间的匹配度;其次,为提取目标区域丰富的细节信息,在原始YOLOv5算法基础上添加频域通道注意力模块FcaNet;最后,采用解耦检测头将分类任务和回归任务分离。在NEU-DET数据集上的实验结果表明,改进的YOLOv5算法在引入较少参数量的情况下,检测精度提高了4.2个百分点,平均精度均值(mAP)达到85.5%,每秒传输帧数(Frames Per Second,FPS)达到27.71,与原YOLOv5相差不大,能够满足检测实时性的要求。
文摘Nowadays, crop diseases are a crucial problem to the world’s food supplies, in a world where the population count is around 7 billion people, with more than 90% not getting access to the use of tools or features that would identify and solve the problem. At present, we live in a world dominated by technology on a significant scale, major network coverage, high-end smart-phones, as long as there are great discoveries and improvements in AI. The combination of high-end smart-phones and computer vision via Deep Learning has made possible what can be defined as “smartphone-assisted disease diagnosis”. In the area of Deep Learning, multiple architecture models have been trained, some achieving performance reaching more than 99.53% [1]. In this study, we evaluate CNN’s architectures applying transfer learning and deep feature extraction. All the features obtained will also be classified by SVM and KNN. Our work is feasible by the use of the open source Plant Village Dataset. The result obtained shows that SVM is the best classifier for leaf’s diseases detection.
基金supported by the National Natural Science Foundations of China(Nos.61961040,61771089)the Sichuan Provincial Key Research and Development Program(No.2021YFQ0011)。
文摘随着低地球轨道(Low Earth orbit,LEO)卫星的网络结构日益复杂,高效的信息传输对于天地一体化网络(Space-ground integrated network,SGIN)的发展至关重要。本文通过同时考虑切换次数、仰角和可用信道数量等指标来优化多目标切换问题。通过数学建模将地球划分为多个区域,并将优化目标定义为切换次数和负载均衡的加权和,根据不同场景确定加权系数。通过设置表示信息传输质量的阈值,可以优化仰角。本文将天地一体化网络的卫星切换问题转化为整数线性规划(Integer linear programming,ILP)问题,并使用数学工具求解,以提供最优解。同时,由于在实际工程应用中基于ILP的策略具有较高的算法复杂性,本文还提出了一种基于二分图的启发式切换策略。通过对一个实际应用的低轨卫星星座(Globalstar)进行仿真实验,验证了本文所提出的切换策略的有效性。
基金supported in part by the Sichuan Provincial Key Research and Development Program (No.2021YFQ0011)the National Natural Science Foundation of China (Nos.61961040, 61771089)。
文摘在智慧城市等应用场景中,单一的无线通信技术无法覆盖复杂的网络环境,且不同的通信技术间无法相互通信,无法充分利用不同的无线通信技术。因此,本文设计了一种可以与多种无线通信节点进行通信的设备,称为无线异构通信模块(Wireless communication heterogeneous modules, WHCM)。WHCM被用作枢纽来构建通信异构网状网络(Communication heterogeneous mesh network, CHMN)。CHMN由具有多种无线通信协议的节点组成。本文提出了一种基于CHMN的按需多路径距离矢量(AD HOC on-demand multi-path distance vector,AOMDV)路由协议(名为CH-AOMDV)。CH-AOMDV能够在路由发起和路径建立过程中识别不同类型的通信协议,并比较CHMN中每个节点的通信距离、数据传输速率、能量和负载情况。NS-2平台仿真表明,当节点之间的分布距离变远时,在数据包传输速率、平均端到端延迟、吞吐率和路由开销方面,CHMN的性能优于传统网络。本文提出了一种实现CHMN的方法和相应的负载均衡路由协议CH-AOMDV,该协议优于其他3种协议,它提高了网络的生命周期、吞吐量和数据包分组投递率,降低了节点开销和平均端到端延迟,该协议对CHMN非常有用。
文摘考虑到无人机群在协同完成任务时对时延的高要求,选用先验式路由协议OLSR(Optimized Link State Routing)协议。但无人机自组网中无人机节点高速移动和能量有限的特性,使得OLSR选举出来的MPR(Multi-Point Relay)节点可能会因此而丧失MPR资格,从而导致时延增加,网络开销增大。针对该问题,提出一种基于节点速度和能量的MPR集选择算法,运用HELLO分组在邻居探测的过程中感知节点能量和速度,之后在MPR选举前根据节点速度和能量对一跳邻居进行预处理,从而使速度快能量低的节点永不成为MPR节点。排除掉节点后,在节点意愿值相同的情况下再次对节点的速度和能量进行加权计算,选出最优MPR节点。仿真结果表明,基于节点速度和能量的MPR集选择算法在时延、吞吐量、节点能量消耗三个指标都具有良好的特性。
