基于深度迁移学习的微型细粒度图像分类

现有的细粒度分类模型不仅利用图像的类别标签,还使用大量人工标注的额外信息。为解决该问题,本文提出一种深度迁移学习模型,将大规模有标签细粒度数据集上学习到的图像特征有效地迁移至微型细粒度数据集中。首先,通过衔接域定量计算域间任务的关联度。然后,根据关联度选择适合目标域的迁移特征。最后,使用细粒度数据集视图类标签进行辅助学习,通过联合学习所有属性来获取更多的特征表示。实验表明,本文方法不仅可以获得较高精度,而且能够有效减少模型训练时间,同时也验证了进行域间特征迁移可以加速网络学习与优化这一结论。     

深度学习在多天气分类算法中的研究与应用

针对目前多天气识别分类问题,提出了一种基于深度学习和计算机视觉的天气现象自动分类算法。采集并建立了一个包括雾霾、沙尘、雨、雪、霜、露6类天气的适用于任意场景的多天气现象数据集,改善了目前已见报数据集规模小、种类单一、只面向特定场景的情况;同时采用密集连接和池化均衡的结构搭建深度卷积神经网络(CNN)模型,训练并挖掘天气数据的特征与内在规律,用深度学习方法实现天气现象的自动分类。实验结果表明:相比传统计算机视觉算法,该算法解决了严重依靠特征提取、适用场景单一问题;且比大多数深度网络模型参数更少、识别准确性更高,算法泛化性能大幅提升。     

基于知识迁移的深度学习无线电信号聚类方法

现有的无线电信号调制识别方法在先验数据不足时通常很难对无类标信号进行有效识别。针对这个问题,本文提出了一种基于知识迁移的深度学习无线电信号聚类方法(DTC)。该方法基于样本对比,分析样本间的相似性,并利用卷积神经网络(CNN)提取无线电信号的特征,同时设计了一种预训练框架,通过迁移同领域数据集的知识,有效提升了CNN特征提取能力,实现了引导聚类方向、提升聚类性能的目标。实验结果表明,该方法在多个公开数据集上的聚类性能都显著优于现有的聚类方法。与现有方法相比,DTC在RML 2016.10A和RML 2016.04C数据集上的聚类精度分别提升了30.34%和28.04%。     

一种基于VGGish神经网络的水声目标识别方法

水声目标智能识别是水声装备智能化的重要组成部分,深度学习则是实现水声目标智能识别的重要技术手段之一。当前水声目标智能识别经常面临数据集较小带来的训练样本量不足的情况,针对小数据集识别中存在的因过拟合导致模型泛化能力不足,以及输入的水声信号二维谱图样式不统一的问题,文章提出了一种基于VGGish神经网络模型的水声目标识别方法。该方法以VGGish网络作为特征提取器,并在VGGish网络前部加入了信号预处理模块,同时设计了一种基于传统机器学习算法的联合分类器,通过以上措施解决了过拟合问题和二维谱图样式不统一问题。实验结果显示,该方法应用在ShipsEar数据集上得到了94.397%的识别准确率,高于传统预训练-微调法得到的最高90.977%的准确率,并且在相同条件下该方法的模型训练耗时仅为传统预训练-微调方法的0.5%左右,有效提高了识别准确率和模型训练速度。     

基于轻量级CNN与主动学习的工件疵病识别方法研究

应用图像识别技术实现工件疵病自动检测可以提高效率,降低人工成本。卷积神经网络(CNN)具有很强的特征提取能力,广泛应用于图像识别等领域。但是,已提出的网络模型普遍存在参数量和计算量巨大,以及海量工件数据集中、人工标注成本高等缺点,很难应用于工件疵病的实时自动识别。本文提出了一种基于轻量级CNN与主动学习的工件疵病识别方法,该方法通过深度可分离卷积和反转残差卷积构建一种轻量级卷积神经网络,并在识别过程中采用主动学习方法不断添加标注样本。实验表明,提出的方法识别精度达到98.3%,并且能节省18.8%的人力标注成本。     

基于ResNet50与迁移学习的轮毂识别

针对人工进行轮毂分拣存在的误识别问题,采用一种基于ResNet50与迁移学习的神经网络模型来识别汽车轮毂。把预训练模型参数迁移到ResNet50卷积神经网络中,修改原网络的输出层,构建基于ResNet50的迁移学习模型,通过进一步训练轮毂数据集来微调模型参数,提取轮毂的细粒度特征。通过对比AlexNet、VGG11、VGG16与ResNet50模型在未使用微调、使用微调和冻结不同数量卷积层参数时的训练效率、准确率,证明ResNet50迁移学模型在冻结前7个Bottleneck残差块参数时不仅能缩短训练时间,并能在相同迭代周期下取得更高的准确率。在该冻结策略下训练生成TL-ResNet50迁移学习模型,分别对8种轮毂进行预测,得出每种轮毂的平均准确率达到99%以上。     

基于光流引导特征的视频行人再识别方法

视频行人再识别的主要任务是基于视频序列判断由不同摄像机捕获的行人是否为同一身份.当前的视频行人再识别方法大都需要提前手动提取光流图来计算行人的运动特征.为了简化网络结构,减少计算量,提出了一种基于光流引导特征的视频行人再识别方法.首先使用卷积神经网络提取行人图像的深度特征,并根据深度特征图计算光流引导特征来描述行人运动信息;然后联合行人空间外观特征与时间运动特征,获得视频级的行人特征描述矢量;最后计算特征描述矢量之间的欧式距离,判定两段图像序列中的行人是否为同一身份.在数据集iLIDS-VID、PRID2011上进行实验,结果表明该方法具有较高的识别率.     

基于通道注意力与迁移学习的红外图像超分辨率重建算法

针对现有红外图像分辨率低、质量不高的问题,提出了基于通道注意力与迁移学习的红外图像超分辨率重建方法.该方法设计了一个深度卷积神经网络,融入通道注意力机制来增强网络的学习能力,并且使用残差学习方式来减轻梯度爆炸或消失问题,加速网络的收敛.考虑到高质量的红外图像难以采集、数目不足的情况,将网络的训练分成两步:第一步使用自然...     

基于深度学习的水体色度分类算法研究

运用深度学习技术进行非接触、快速水体色度检测与分类,采用无人机采集水体图像,运用色度仪对标定的图像完成分类,建立数据集。采用图像归一化处理减少环境因素对分类结果的影响,设计多特征的分步边缘检测算法,检测水域图像边缘,剔除无关像素。对VGG 16、GoogleNet-V3和ResNet 18卷积神经网络进行水体色度分类模型构建与训练,后筛选Inception结构和残差结构为基本构建单元,设计专门用于水体色度分类的WCNet 15与WCNet 21神经网络模型。在训练集上训练参数并利用验证集完成对2个模型的准确率的比较,筛选准确率高的WCNet 21模型作为最终水体色度分类模型。WCNet21模型的最优准确率可达97.8%,满足水体色度分类需求,可应用到具体的水体色度分类工作当中。     

基于迁移学习的异步电机故障诊断

针对异步电机故障诊断中,故障数据样本少导致传统深度神经网络模型泛化能力差的问题,提出一种异构迁移学习的异步电机故障诊断算法。首先,通过仿真平台模拟异步电机故障,以解决故障数据样本少的问题;其次,对正常和故障状态下的电流电压信号进行小波变换,作为深度学习网络的输入;然后,基于多核最大平均差异方法,获得仿真数据和实测数据的深度特征差异,对深度学习神经网络参数微调,使其深度学习特征具有跨域不变性。最终,在实验平台上验证文中所提算法,实验结果表明,该算法的故障诊断准确率高,依赖实测故障数据样本少。     

基于深度学习的烟包识别与分类

目的 提取烟包图像数据训练深度学习目标检测模型,提升烟包流水线拣包效率和准确性。方法 基于深度学习建立一种烟包识别分类模型,对原始YOLOv3模型进行改进,在原网络中加入设计的多空间金字塔池化结构(M–SPP),将64×64尺度的特征图下采样与32×32尺度的特征图进行拼接,并去除16×16尺度的预测特征层,提高模型的检测准确率和速度,并采用K–means++算法对先验框参数进行优化。结果 实验表明该目标检测模型平均准确率达到99.68%,检测速度达到70.82帧/s。结论 基于深度学习建立的图像识别分类模型准确率高且检测速度快,有效满足烟包流水线自动化实时检测。     

基于深度学习的地震子波分类识别

针对地震勘探中噪声压制的问题,构建了一种适合分类和识别地震子波的卷积神经网络模型.首先对卷积神经网络模型的激活函数、卷积核大小以及归一化层等进行了设计,然后利用已搭建好的卷积神经网络对地震信号的时频谱图进行特征提取,最后实现了不同类型的含噪地震信号的分类和识别.实验结果表明,该模型有高分类率和识别率及较好的抗干扰能力,...     

基于混合迁移学习的运动想象分类算法研究及其在脑机接口中的应用

为提升迁移学习在运动想象脑机接口应用过程中的迁移高效性及普适性,综合实例迁移和特征迁移学习方法的优势进而构建了混合迁移学习模型。首先,依据样本权重极化原理改进TrAdaBoost算法以实现实例层面的迁移,优化源域训练样本;其次,基于大间隔投射迁移支持向量机进一步缩短源域与目标域间的分布距离以完成特征层面的迁移,实现迁移效率最大化。进一步,将该方法应用于脑机接口竞赛Dataset IIb数据集进行离线测试及分析,研究结果表明混合迁移学习模型的迁移效率明显高于单一迁移学习模型,并且对于不同迁移对象识别准确率相对提升均值在70%以上,验证了所述方法的有效性与普适性。此外,基于已搭建的运动想象识别系统进行在线测试,验证了模型的实用性。     

Deep Transfer Learning Based Detection and Classification of Citrus Plant Diseases

Citrus fruit crops are among the world’s most important agricultural products, but pests and diseases impact their cultivation, resulting in yield and quality losses. Computer vision and machine learning have been widely used to detect and classify plant diseases over the last decade, allowing for early disease detection and improving agricultural production. This paper presented an automatic system for the early detection and classification of citrus plant diseases based on a deep learning (DL) model, which improved accuracy while decreasing computational complexity. The most recent transfer learning-based models were applied to the Citrus Plant Dataset to improve classification accuracy. Using transfer learning, this study successfully proposed a Convolutional Neural Network (CNN)-based pre-trained model (EfficientNetB3, ResNet50, MobiNetV2, and InceptionV3) for the identification and categorization of citrus plant diseases. To evaluate the architecture’s performance, this study discovered that transferring an EfficientNetb3 model resulted in the highest training, validating, and testing accuracies, which were 99.43%, 99.48%, and 99.58%, respectively. In identifying and categorizing citrus plant diseases, the proposed CNN model outperforms other cutting-edge CNN model architectures developed previously in the literature.     

Sentiment Analysis Method Based on Kmeans and Online Transfer Learning

Sentiment analysis is a research hot spot in the field of natural language processing and content security. Traditional methods are often difficult to handle the problems of large difference in sample distribution and the data in the target domain is transmitted in a streaming fashion. This paper proposes a sentiment analysis method based on Kmeans and online transfer learning in the view of fact that most existing sentiment analysis methods are based on transfer learning and offline transfer learning. We first use the Kmeans clustering algorithm to process data from one or multiple source domains and select the data similar to target domain data to establish the classifier, so that the processed data does not negatively transfer the data in the target domain. And then create a new classifier based on the new target domain. The source domain classifier and target domain classifier are combined with certain weights by using the homogeneous online transfer learning method to achieve sentiment analysis. The experimental results show that this method has achieved better performance in terms of error rate and classification accuracy.     

Classification and Research of Skin Lesions Based on Machine Learning

Classification of skin lesions is a complex identification challenge. Due to the wide variety of skin lesions, doctors need to spend a lot of time and effort to judge the lesion image which zoomed through the dermatoscopy. The diagnosis which the algorithm of identifying pathological images assists doctors gets more and more attention. With the development of deep learning, the field of image recognition has made longterm progress. The effect of recognizing images through convolutional neural network models is better than traditional image recognition technology. In this work, we try to classify seven kinds of lesion images by various models and methods of deep learning, common models of convolutional neural network in the field of image classification include ResNet, DenseNet and SENet, etc. We use a fine-tuning model with a multi-layer perceptron, by training the skin lesion model, in the validation set and test set we use data expansion based on multiple cropping, and use five models’ ensemble as the final results. The experimental results show that the program has good results in improving the sensitivity of skin lesion diagnosis.     

Multi-View & Transfer Learning for Epilepsy Recognition Based on EEG Signals

Epilepsy is a central nervous system disorder in which brain activity becomes abnormal. Electroencephalogram (EEG) signals, as recordings of brain activity, have been widely used for epilepsy recognition. To study epileptic EEG signals and develop artificial intelligence (AI)-assist recognition, a multi-view transfer learning (MVTL-LSR) algorithm based on least squares regression is proposed in this study. Compared with most existing multi-view transfer learning algorithms, MVTL-LSR has two merits: (1) Since traditional transfer learning algorithms leverage knowledge from different sources, which poses a significant risk to data privacy. Therefore, we develop a knowledge transfer mechanism that can protect the security of source domain data while guaranteeing performance. (2) When utilizing multi-view data, we embed view weighting and manifold regularization into the transfer framework to measure the views’ strengths and weaknesses and improve generalization ability. In the experimental studies, 12 different simulated multi-view & transfer scenarios are constructed from epileptic EEG signals licensed and provided by the University of Bonn, Germany. Extensive experimental results show that MVTL-LSR outperforms baselines. The source code will be available on .     

Cr9Mo铸钢组织和性能的研究

研究了Cr9Mo铸钢组织和性能之间的关系。依据试验结果,指出了热处理工艺变化对其组织、性能的影响,并与Cr5Mo铸钢作比较。     

A Novel Scene Text Recognition Method Based on Deep Learning

Scene text recognition is one of the most important techniques in pattern recognition and machine intelligence due to its numerous practical applications. Scene text recognition is also a sequence model task. Recurrent neural network (RNN) is commonly regarded as the default starting point for sequential models. Due to the non-parallel prediction and the gradient disappearance problem, the performance of the RNN is difficult to improve substantially. In this paper, a new TRDD network architecture which base on dilated convolution and residual block is proposed, using Convolutional Neural Networks (CNN) instead of RNN realizes the recognition task of sequence texts. Our model has the following three advantages in comparison to existing scene text recognition methods: First, the text recognition speed of the TRDD network is much fast than the state-of-the-art scene text recognition network based recurrent neural networks (RNN). Second, TRDD is easier to train, avoiding the problem of exploding and vanishing, which is major issue for RNN. Third, both using larger dilated factors and increasing the filter size are all viable ways to change receptive field size. We benchmark the TRDD on four standard datasets, it has higher recognition accuracy and faster recognition speed based on the smaller model. It is hopefully used in the real-time application.