Diagnosis of Middle Ear Diseases Based on Convolutional Neural Network

An otoscope is traditionally used to examine the eardrum and ear canal.A diagnosis of otitis media(OM)relies on the experience of clinicians.If an examiner lacks experience,the examination may be difficult and time-consuming.This paper presents an ear disease classification method using middle ear images based on a convolutional neural network(CNN).Especially the segmentation and classification networks are used to classify an otoscopic image into six classes:normal,acute otitis media(AOM),otitis media with effusion(OME),chronic otitis media(COM),congenital cholesteatoma(CC)and traumatic perforations(TMPs).The Mask R-CNN is utilized for the segmentation network to extract the region of interest(ROI)from otoscopic images.The extracted ROIs are used as guiding features for the classification.The classification is based on transfer learning with an ensemble of two CNN classifiers:EfficientNetB0 and Inception-V3.The proposed model was trained with a 5-fold cross-validation technique.The proposed method was evaluated and achieved a classification accuracy of 97.29%.

Hybrid Convolutional Neural Network for Plant Diseases Prediction

Plant diseases prediction is the essential technique to prevent the yield loss and gain high production of agricultural products.The monitoring of plant health continuously and detecting the diseases is a significant for sustainable agri-culture.Manual system to monitor the diseases in plant is time consuming and report a lot of errors.There is high demand for technology to detect the plant dis-eases automatically.Recently image processing approach and deep learning approach are highly invited in detection of plant diseases.The diseases like late blight,bacterial spots,spots on Septoria leaf and yellow leaf curved are widely found in plants.These are the main reasons to affects the plants life and yield.To identify the diseases earliest,our research presents the hybrid method by com-bining the region based convolutional neural network(RCNN)and region based fully convolutional networks(RFCN)for classifying the diseases.First the leaf images of plants are collected and preprocessed to remove noisy data in image.Further data normalization,augmentation and removal of background noises are done.The images are divided as testing and training,training images are fed as input to deep learning architecture.First,we identify the region of interest(RoI)by using selective search.In every region,feature of convolutional neural network(CNN)is extracted independently for further classification.The plants such as tomato,potato and bell pepper are taken for this experiment.The plant input image is analyzed and classify as healthy plant or unhealthy plant.If the image is detected as unhealthy,then type of diseases the plant is affected will be displayed.Our proposed technique achieves 98.5%of accuracy in predicting the plant diseases.

Grid Side Distributed Energy Storage Cloud Group End Region Hierarchical Time-Sharing Configuration Algorithm Based onMulti-Scale and Multi Feature Convolution Neural Network

There is instability in the distributed energy storage cloud group end region on the power grid side.In order to avoid large-scale fluctuating charging and discharging in the power grid environment and make the capacitor components showa continuous and stable charging and discharging state,a hierarchical time-sharing configuration algorithm of distributed energy storage cloud group end region on the power grid side based on multi-scale and multi feature convolution neural network is proposed.Firstly,a voltage stability analysis model based onmulti-scale and multi feature convolution neural network is constructed,and the multi-scale and multi feature convolution neural network is optimized based on Self-OrganizingMaps(SOM)algorithm to analyze the voltage stability of the cloud group end region of distributed energy storage on the grid side under the framework of credibility.According to the optimal scheduling objectives and network size,the distributed robust optimal configuration control model is solved under the framework of coordinated optimal scheduling at multiple time scales;Finally,the time series characteristics of regional power grid load and distributed generation are analyzed.According to the regional hierarchical time-sharing configuration model of“cloud”,“group”and“end”layer,the grid side distributed energy storage cloud group end regional hierarchical time-sharing configuration algorithm is realized.The experimental results show that after applying this algorithm,the best grid side distributed energy storage configuration scheme can be determined,and the stability of grid side distributed energy storage cloud group end region layered timesharing configuration can be improved.

Ozone Depletion Identification in Stratosphere Through Faster Region-Based Convolutional Neural Network

The concept of classification through deep learning is to build a model that skillfully separates closely-related images dataset into different classes because of diminutive but continuous variations that took place in physical systems over time and effect substantially.This study has made ozone depletion identification through classification using Faster Region-Based Convolutional Neural Network(F-RCNN).The main advantage of F-RCNN is to accumulate the bounding boxes on images to differentiate the depleted and non-depleted regions.Furthermore,image classification’s primary goal is to accurately predict each minutely varied case’s targeted classes in the dataset based on ozone saturation.The permanent changes in climate are of serious concern.The leading causes beyond these destructive variations are ozone layer depletion,greenhouse gas release,deforestation,pollution,water resources contamination,and UV radiation.This research focuses on the prediction by identifying the ozone layer depletion because it causes many health issues,e.g.,skin cancer,damage to marine life,crops damage,and impacts on living being’s immune systems.We have tried to classify the ozone images dataset into two major classes,depleted and non-depleted regions,to extract the required persuading features through F-RCNN.Furthermore,CNN has been used for feature extraction in the existing literature,and those extricated diverse RoIs are passed on to the CNN for grouping purposes.It is difficult to manage and differentiate those RoIs after grouping that negatively affects the gathered results.The classification outcomes through F-RCNN approach are proficient and demonstrate that general accuracy lies between 91%to 93%in identifying climate variation through ozone concentration classification,whether the region in the image under consideration is depleted or non-depleted.Our proposed model presented 93%accuracy,and it outperforms the prevailing techniques.

基于改进Faster R-CNN与U-Net算法的桥梁病害识别与量化方法

为实现桥梁病害检测的自动化,对基于图像处理技术的混凝土桥梁表观病害的智能识别和尺寸确定方法展开研究.提出基于改进Faster R-CNN算法的病害识别方法,利用K均值聚类和遗传算法对区域候选网络锚框进行优化设计;以裂缝预测区域为基础,提出ResNet34结合U-Net的裂缝形态提取方法,并结合裂缝形态学研究了裂缝像素宽度和长度的确定方法.结果表明:锚框优化设计可改进Faster R-CNN算法的表观病害识别效果,5类常见病害的预测准确率、召回率、平均精确率分别由68.40%、69.87%、74.64%提升到85.40%、83.59%、83.72%;利用病害预测框,结合改进U-Net算法的裂缝像素尺寸计算,可实现裂缝病害尺寸的自动测量;基于改进Faster R-CNN和改进U-Net的方法可实现混凝土桥梁常见病害的智能识别和尺寸量化,从而提高桥梁病害检测效率并促进桥梁技术状况评定的智能化.

改进Mask R-CNN的无人机影像建筑物提取

从无人机影像中自动提取建筑物对城乡规划和管理至关重要,然而,在复杂背景干扰和建筑物外观变化很大的情况下给实例提取带来挑战。因此,提出一种改进的Mask区域卷积神经网络(R-CNN)方法用于无人机影像的建筑物自动实例提取。改进方法以ResNet-101作为特征提取网络,在特征融合网络方面,通过添加自底向上的路径增强整个特征层次的定位能力,同时在特征融合中加入空洞空间金字塔池化模块(ASPP)来提高多尺度能力与改善模型性能。在自制建筑物数据集上的综合实验结果表明,与原始的Mask R-CNN方法相比,改进方法的mAP值提高了2.6%,能够很好地实现无人机影像建筑物实例提取。

Facial Expression Recognition Using Enhanced Convolution Neural Network with Attention Mechanism

Facial Expression Recognition(FER)has been an interesting area of research in places where there is human-computer interaction.Human psychol-ogy,emotions and behaviors can be analyzed in FER.Classifiers used in FER have been perfect on normal faces but have been found to be constrained in occluded faces.Recently,Deep Learning Techniques(DLT)have gained popular-ity in applications of real-world problems including recognition of human emo-tions.The human face reflects emotional states and human intentions.An expression is the most natural and powerful way of communicating non-verbally.Systems which form communications between the two are termed Human Machine Interaction(HMI)systems.FER can improve HMI systems as human expressions convey useful information to an observer.This paper proposes a FER scheme called EECNN(Enhanced Convolution Neural Network with Atten-tion mechanism)to recognize seven types of human emotions with satisfying results in its experiments.Proposed EECNN achieved 89.8%accuracy in classi-fying the images.

Leguminous seeds detection based on convolutional neural networks:Comparison of Faster R-CNN and YOLOv4 on a small custom dataset

This paper help with leguminous seeds detection and smart farming. There are hundreds of kinds of seeds and itcan be very difficult to distinguish between them. Botanists and those who study plants, however, can identifythe type of seed at a glance. As far as we know, this is the first work to consider leguminous seeds images withdifferent backgrounds and different sizes and crowding. Machine learning is used to automatically classify andlocate 11 different seed types. We chose Leguminous seeds from 11 types to be the objects of this study. Thosetypes are of different colors, sizes, and shapes to add variety and complexity to our research. The images datasetof the leguminous seeds was manually collected, annotated, and then split randomly into three sub-datasetstrain, validation, and test (predictions), with a ratio of 80%, 10%, and 10% respectively. The images consideredthe variability between different leguminous seed types. The images were captured on five different backgrounds: white A4 paper, black pad, dark blue pad, dark green pad, and green pad. Different heights and shootingangles were considered. The crowdedness of the seeds also varied randomly between 1 and 50 seeds per image.Different combinations and arrangements between the 11 types were considered. Two different image-capturingdevices were used: a SAMSUNG smartphone camera and a Canon digital camera. A total of 828 images wereobtained, including 9801 seed objects (labels). The dataset contained images of different backgrounds, heights,angles, crowdedness, arrangements, and combinations. The TensorFlow framework was used to construct theFaster Region-based Convolutional Neural Network (R-CNN) model and CSPDarknet53 is used as the backbonefor YOLOv4 based on DenseNet designed to connect layers in convolutional neural. Using the transfer learningmethod, we optimized the seed detection models. The currently dominant object detection methods, Faster RCNN, and YOLOv4 performances were compared experimentally. The mAP (mean average precision) of the FasterR-CNN and YOLOv4 models were 84.56% and 98.52% respectively. YOLOv4 had a significant advantage in detection speed over Faster R-CNN which makes it suitable for real-time identification as well where high accuracy andlow false positives are needed. The results showed that YOLOv4 had better accuracy, and detection ability, as wellas faster detection speed beating Faster R-CNN by a large margin. The model can be effectively applied under avariety of backgrounds, image sizes, seed sizes, shooting angles, and shooting heights, as well as different levelsof seed crowding. It constitutes an effective and efficient method for detecting different leguminous seeds incomplex scenarios. This study provides a reference for further seed testing and enumeration applications.

基于Faster R-CNN的密集人群检测算法

为提高拥挤场景下的人群检测准确率,提出一种基于改进Faster R-CNN的密集人群检测算法。首先,在特征提取阶段添加空间与通道注意力机制,使用加强的双向特征金字塔网络(S-BiFPN)替代原网络中的多尺度特征金字塔(FPN),使网络对重要特征进行自主学习并加强对图像深层特征的提取;其次,引入多实例预测(MIP)算法对实例进行预测,以避免模型对拥挤场景下的目标造成漏检;最后,对模型中的非极大值抑制(NMS)进行优化,并额外增设一个交并比(IoU)阈值,以对检测结果的干扰项进行精确抑制。在开源的密集人群检测数据集上进行测试的结果显示,相较于原Faster R-CNN算法,所提算法的平均精度(AP)提升5.6%,Jaccard指数值提升3.2%。所提算法具有较高检测精度和稳定性,可以满足密集场景人群检测的需求。

基于Mask R-CNN的柑橘主叶脉显微图像实例分割模型

针对目前植物解剖表型的测量与分析过程自动化低,难以应对复杂解剖表型的提取和识别的问题,以柑橘主叶脉为研究对象,提出了一种基于掩膜区域卷积神经网络(Mask region convolutional neural network,Mask R-CNN)的主叶脉显微图像实例分割模型,以残差网络ResNet50和特征金字塔(Feature pyramid network,FPN)为主干特征提取网络,在掩膜(Mask)分支上添加一个新的感兴趣区域对齐层(Region of interest Align,RoI-Align),提升Mask分支的分割精度。结果表明,该网络架构能够精准地对柑橘主叶脉横切面中的髓部、木质部、韧皮部和皮层细胞进行识别分割。Mask R-CNN模型对髓部、木质部、韧皮部和皮层细胞的分割平均精确率(交并比(IoU)为0.50)分别为98.9%、89.8%、95.7%和97.2%,对4个组织区域的分割平均精确率均值(IoU为0.50)为95.4%。与未在Mask分支添加RoI-Align的Mask R-CNN相比,精度提升1.6个百分点。研究结果表明,Mask R-CNN模型对柑橘主叶脉各类组织区域具有良好的识别分割效果,可为柑橘微观表型研究提供技术支持与研究基础。

基于改进Faster R-CNN算法的行人识别系统设计与研究

文章基于改进更快的区域卷积神经网络(Faster Region Convolutional Neural Networks,Faster R-CNN)模型,提出了一种行人识别系统设计。介绍了计算机视觉常用技术手段与方法、通行检测步骤,分析了主流的算法优缺点,利用深度学习方法提取图像特征,然后使用改进Faster R-CNN模型进行目标检测。在改进Faster R-CNN模型中,采用了自适应尺度池化和增强的感兴趣区域(Region of Interest,RoI)池化技术,可以提高模型检测精度和速度。

基于Faster R-CNN的人脸面部情感识别方法

常规人脸面部情感识别方法不准确,存在识别后的情感反馈误差大的问题,为此提出基于更快的区域卷积神经网络(Faster Region-Convolutional Neural Network,Faster R-CNN)的人脸面部情感识别方法。首先,采集人脸图像数据,通过面部检测、面部对齐、面部数据增强、面部归一化4个步骤预处理面部图像;其次,基于多尺度特征融合算法提取表情特征,生成情感识别数据标签;最后,利用FasterR-CNN构建人脸面部情感识别模型,并识别人脸面部情感。实验结果表明,基于FasterR-CNN的人脸面部情感识别方法在6种基本表情中均具有90%以上的识别准确率。

基于Faster R-CNN的海底管道智能检测方法

为提高海底管道缺陷及组件的检测精度并实现智能化海底管道安全检测,提出一种基于快速区域卷积神经网络(Faster R-CNN)的海底管道智能检测方法。首先,通过基值校正和分段映射-伪彩色化方法,将漏磁检测信号转化为伪彩色图,以增强漏磁信号的关键特征;其次,基于多模态数据增强来提升检测模型的泛化能力;然后,基于多模态数据增强后的样本训练改进的Faster R-CNN网络,建立最优的智能检测模型;最后,以试验场和渤海在役管道为例,验证所提方法的有效性。结果表明:所提方法的平均检测精度可达93.8%,相较原始的Faster R-CNN算法提高8%,且平均交并比达到0.75,能够精准地实现海底油气管道多目标检测,保障海底管道的安全运行。

Unconstrained Gender Recognition from Periocular Region Using Multiscale Deep Features

The gender recognition problem has attracted the attention of the computer vision community due to its importance in many applications(e.g.,sur-veillance and human–computer interaction[HCI]).Images of varying levels of illumination,occlusion,and other factors are captured in uncontrolled environ-ments.Iris and facial recognition technology cannot be used on these images because iris texture is unclear in these instances,and faces may be covered by a scarf,hijab,or mask due to the COVID-19 pandemic.The periocular region is a reliable source of information because it features rich discriminative biometric features.However,most existing gender classification approaches have been designed based on hand-engineered features or validated in controlled environ-ments.Motivated by the superior performance of deep learning,we proposed a new method,PeriGender,inspired by the design principles of the ResNet and DenseNet models,that can classify gender using features from the periocular region.The proposed system utilizes a dense concept in a residual model.Through skip connections,it reuses features on different scales to strengthen dis-criminative features.Evaluations of the proposed system on challenging datasets indicated that it outperformed state-of-the-art methods.It achieved 87.37%,94.90%,94.14%,99.14%,and 95.17%accuracy on the GROUPS,UFPR-Periocular,Ethnic-Ocular,IMP,and UBIPr datasets,respectively,in the open-world(OW)protocol.It further achieved 97.57%and 93.20%accuracy for adult periocular images from the GROUPS dataset in the closed-world(CW)and OW protocols,respectively.The results showed that the middle region between the eyes plays a crucial role in the recognition of masculine features,and feminine features can be identified through the eyebrow,upper eyelids,and corners of the eyes.Furthermore,using a whole region without cropping enhances PeriGender’s learning capability,improving its understanding of both eyes’global structure without discontinuity.

多区域注意力的细粒度图像分类网络

目前细粒度图像分类的难点在于如何精准定位图像中高度可辨的局部区域以及其他辅助判别特征。提出一种多区域注意力的细粒度图像分类网络来解决这个问题。首先使用Inception-V3对图像特征进行提取,通过重复使用注意力擦除的方法使模型关注次要特征;然后通过背景去除以及上采样的方法获取图像更精准的局部图像,对提取到的局部特征进行位置统计,并以矩形框的方式获取图像整体,减少细节信息丢失;最后对局部与整体图像进行更加细致的学习。此外,设计联合损失函数,通过动态平衡难易样本和缩小类内差距的方法改善模型的识别效果。实验结果表明,该方法在公开的细粒度图像数据集CUB-200-2011、Stanford-Cars和FGVC-Aircraft上的准确率分别达到89.2%、94.8%、94.0%,相较于对比方法性能更优。

基于深层图卷积的EEG情绪识别方法研究

针对浅层图卷积提取的局部脑区空间关联信息对情感脑电表征不足的问题,本文提出了一种深层图卷积网络模型。该模型利用深层图卷积学习情绪脑电全局通道间的内在关系,在卷积传播过程中应用残差连接和权重自映射解决深层图卷积网络面临的节点特征收敛到固定空间无法学习到有效特征的问题,并在卷积层后加入PN正则化扩大不同情绪特征间的距离,提高情绪识别的性能。在SEED数据集上进行实验,与浅层图卷积网络相比准确率提高了0.7%,标准差下降了3.15。结果表明该模型提取的全局脑区空间关联信息对情绪识别的有效性。

基于CNN-OBIA的黄河源区水体提取及时空变化

准确识别水体信息是分析地表水时空动态变化的重要技术手段。针对目前各种长时序水体信息提取方法精度低的问题,基于Landsat遥感影像,选用1986~2022年5484景黄河源区遥感影像,分别运用卷积神经网络结合面向对象(CNN-OBIA)和多指数水体检测规则(MIWDR)两种方法提取了黄河源区的地表水体,并对两种方法的提取精度进行了对比分析。在此基础上,探究了1986~2022年黄河源区水体信息的时空变化特征,并对其主要气候因素进行相关分析。结果表明:①CNN-OBIA的总体精度和Kappa系数分别为96.78%和0.93,MIWDR的总体精度和Kappa系数分别为94.28%和0.88,总体而言,CNN-OBIA的提取精度高于MIWDR方法。CNN-OBIA的提取结果可以很好地保持水体边界完整性和有效去除山体阴影,可以较好地对细小河流进行提取。②研究区水体总面积呈现出先减少(1986~2001年)后增加(2001~2022年)的变化趋势。③相关性分析表明,降水和气温与水体面积的变化均表现出显著正相关。

多尺度特征和极化自注意力的Faster-RCNN水漂垃圾识别

针对小目标水漂垃圾形态多变、分辨率低且信息有限,导致检测效果不理想的问题,提出一种改进的Faster-RCNN(Faster Regions with Convolutional Neural Network)水漂垃圾检测算法MP-Faster-RCNN(Faster-RCNN with Multi-scale feature and Polarized self-attention)。首先,建立黄河兰州段小目标水漂垃圾数据集,将空洞卷积结合ResNet-50代替原来的VGG-16(Visual Geometry Group 16)作为主干特征提取网络,扩大感受野以提取更多小目标特征;其次,在区域生成网络(RPN)利用多尺度特征,设置3×3和1×1的两层卷积,补偿单一滑动窗口造成的特征丢失;最后,在RPN前加入极化自注意力,进一步利用多尺度和通道特征提取更细粒度的多尺度空间信息和通道间依赖关系,生成具有全局特征的特征图,实现更精确的目标框定位。实验结果表明,MP-Faster-RCNN能有效提高水漂垃圾检测精度,与原始Faster-RCNN相比,平均精度均值(mAP)提高了6.37个百分点,模型大小从521 MB降到了108 MB,且在同一训练批次下收敛更快。

改进卷积神经网络的医学图像感兴趣区域识别

图像中的噪声会提高图像特征信息提取难度,影响图像识别时的细节保留效果,为此提出改进卷积神经网络的医学图像感兴趣区域识别方法。分析医学图像主要噪声来源,构建噪声模型,利用非局部均值滤波算法计算图像全部像素的加权平均值,完成图像去噪处理;通过图像求反、对比度增加和灰度调节等操作增强图像细节信息;利用局部区域特征提取方法获取图像基础纹理特征,包括灰度、平滑度与熵值等;建立具有卷积层、池化层、全连接层的卷积神经网络模型,引入区域建议网络对其改进,通过该网络确定识别的候选区域,将图像特征作为网络输入,经过不断学习迭代,输出最终感兴趣区域。实验结果表明,所提方法在提高图像质量的基础上,识别出的感兴趣区域较为完整,包含的有用信息更多。

基于RMT-CNN的电网短路故障定位研究

随着我国智能电网的快速发展,电网监测数据呈现多元化、高速化、海量化的趋势.为了充分挖掘电力大数据的潜在价值,实现电网内异常区域的自动识别与定位,本文研究了基于随机矩阵理论(random matrix theory,RMT)和卷积神经网络(convolutional neural networks,CNN)的电网异常事件定位方法.首先根据电网内部联系将电网划分为若干子系统,分区构建监测矩阵;然后采用RMT作为数据挖掘的特征提取方法,提取分区矩阵特征向量作为输入,根据电网监测数据和异常识别需求的特点搭建CNN模型;最后基于分区矩阵特征向量构建数据集,训练获得有效的异常事件自动定位CNN模型.以IEEE39节点电网模型三相短路故障为例,分析表明通过RMT提取特征向量的预处理方法能有效降低数据维度,提高CNN模型的故障定位准确率,分区RMT-CNN模型能有效定位电网内异常事件的发生地点,定位精度可达97.96%,精确率可达98.65%.