检索结果-南通市图书馆

Video summarization via global feature difference optimization

在线全文

学校读者我要写书评

暂无评论

Optoelectronics Letters 2023年第9期19卷 570-576页

作者： zhang yunzuo LIU Yameng School of Information Science and Technology Shijiazhuang Tiedao UniversityShijiazhuang 050043China

Video summarization aims at selecting valuable clips for browsing videos with high efficiency.Previous approaches typically focus on aggregating temporal features while ignoring the potential role of visual representations in summarizing videos.In this paper,we present a global difference-aware network(GDANet)that exploits the feature difference across frame and video as guidance to enhance visual features.Initially,a difference optimization module(DOM)is devised to enhance the discriminability of visual features,bringing gains in accurately aggregating temporal cues.Subsequently,a dual-scale attention module(DSAM)is introduced to capture informative contextual information.Eventually,we design an adaptive feature fusion module(AFFM)to make the network adaptively learn context representations and perform feature fusion effectively.We have conducted experiments on benchmark datasets,and the empirical results demonstrate the effectiveness of the proposed framework.

关键词： Video summarization via global feature difference optimization

Computer Vision-Based Human Body Posture Correction System

在线全文

学校读者我要写书评

暂无评论

Mechanical Engineering Science 2024年第1期6卷 1-7页

作者： Yangsen QIU Yukun WANG Yuchen WU Xinyi QIANG yunzuo zhang School of Information science and Technology Shijiazhuang Tiedao UniversityShijiazhuangHebei050043China

With the development of technology and the progress of life,more and more people,regardless of entertainment,learning,or work,cannot do without computer desks and cannot put down their mobile phones.Due to prolonged sitting and often neglecting the importance of posture,incorrect posture can often lead to health problems such as hunchback,lumbar muscle strain,and shoulder and neck pain over time.To address this issue,we designed a computer vision-based human body posture detection system.The system utilizes YOLOv8 technology to accurately locate key points of the human body skeleton,and then analyzes the coordinate positions and depth information of these key points to establish a criterion for distinguishing different postures.With the assistance of an SVM classifier,the system achieves an average recognition rate of 95%.Finally,we successfully deployed the posture detection system on Raspberry Pi hardware and conducted extensive testing.The test results demonstrate that the system can effectively detect various postures and provide real-time reminders to users to correct poor posture,demonstrating good practicality and stability.

关键词： computer vision human posture deep learning image processing

Key Frame Extraction Algorithm of Surveillance Video Based on Quaternion Fourier Significance Detection

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Journal of New Media 2022年第1期4卷 1-11页

作者： zhang yunzuo zhang Jiayu Cai Zhaoquan School of Information Science and Technology Shijiazhuang Tiedao UniversityShijiazhuang050043HebeiChina Shanwei Institute of Technology Shanwei516600GuangdongChina

With the improvement of people’s security awareness,numerous monitoring equipment has been put into use,resulting in the explosive growth of surveillance video data.Key frame extraction technology is a paramount technology for improving video storage efficiency and enhancing the accuracy of video retrieval.It can extract key frame sets that can express video content from massive videos.However,the existing key frame extraction algorithms of surveillance video still have deficiencies,such as the destruction of image information integrity and the inability to extract key frames accurately.To this end,this paper proposes a key frame extraction algorithm of surveillance video based on quaternion Fourier saliency detection.Firstly,the algorithm used colors,and intensity features to perform quaternion Fourier transform on surveillance video sequences.Next,the phase spectrum of the quaternion Fourier transformed image was obtained,and he image visual saliency map was obtained according to the quaternion Fourier phase spectrum.Then,the image visual saliency map of two adjacent frames is used to characterize the change of target motion state.Finally,the frames that can accurately express the motion state of the target are selected as key frames.The experimental results show that the method proposed in this paper can accurately capture the changes of the local motion state of the target while maintaining the integrity of the image information.

关键词： Quaternion fourier transform phase spectrum image saliency map motion status

融合多尺度信息和跨维特征引导的轻量行人检测

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

光电子．激光 2024年第4期35卷 344-350页

作者：张云佐李文博郭威石家庄铁道大学信息科学与技术学院河北石家庄050043 河北省电磁环境效应与信息处理重点实验室河北石家庄050043

针对复杂道路场景下行人检测精度与速度难以提升的问题,提出一种融合多尺度信息和跨维特征引导的轻量级行人检测算法。首先以高性能检测器YOLOX为基础框架,构建多尺度轻量卷积并嵌入主干网络中,以获取多尺度特征信息。然后设计了一种端... 详细信息

针对复杂道路场景下行人检测精度与速度难以提升的问题,提出一种融合多尺度信息和跨维特征引导的轻量级行人检测算法。首先以高性能检测器YOLOX为基础框架,构建多尺度轻量卷积并嵌入主干网络中,以获取多尺度特征信息。然后设计了一种端到端的轻量特征引导注意力模块,采用跨维通道加权的方式将空间信息与通道信息融合,引导模型关注行人的可视区域。最后为减少模型在轻量化过程中特征信息的损失,使用增大感受野的深度可分离卷积构建特征融合网络。实验结果表明,相比于其他主流检测算法,所提算法在KITTI数据集上达到了71.03%的检测精度和80 FPS的检测速度,在背景复杂、密集遮挡、尺度不一等场景中都具有较好的鲁棒性和实时性。

关键词：行人检测多尺度跨维特征引导特征融合轻量化模型

在线全文

学校读者我要写书评

暂无评论

无锚框的轻量级遥感图像目标检测算法

南京信息工程大学学报 2024年第2期16卷 212-220页

作者：张云佐武存宇郭威赵宁石家庄铁道大学信息科学与技术学院石家庄050043 石家庄铁道大学河北省电磁环境效应与信息处理重点实验室石家庄050043 石家庄铁道大学管理学院石家庄050043

现有遥感图像目标检测算法存在参数量大、检测速度慢和难以部署于移动设备的问题,为此,本文提出了一种无锚框的轻量级遥感图像目标检测算法.首先设计了DWS-Sandglass轻量化模块以降低模型体积,并改进模型激活函数,以确保检测精度.然后... 详细信息

现有遥感图像目标检测算法存在参数量大、检测速度慢和难以部署于移动设备的问题,为此,本文提出了一种无锚框的轻量级遥感图像目标检测算法.首先设计了DWS-Sandglass轻量化模块以降低模型体积,并改进模型激活函数,以确保检测精度.然后引入无参数注意力模块SimAM,使网络能够专注于更重要的特征信息.最后对无锚框算法的冗余通道进行剪枝操作以减少模型参数量,并通过微调回升精度.在HRSC2016数据集上的实验结果表明,与当前主流的无锚框检测算法相比,该算法在检测精度相当的情况下检测速度更快、模型体积更小,更适合在移动设备部署.

关键词：计算机应用遥感目标检测轻量级模型剪枝

Power Plant Indicator Light Detection System Based on Improved YOLOv5

在线全文

学校读者我要写书评

暂无评论

Journal of Beijing Institute of Technology 2022年第6期31卷 605-612页

作者： yunzuo zhang Kaina Guo School of Information Science and Technology Shijiazhuang Tiedao UniversityShijiazhuang 050043China

Electricity plays a vital role in daily life and economic development.The status of the indicator lights of the power plant needs to be checked regularly to ensure the normal supply of electricity.Aiming at the problem of a large amount of data and different sizes of indicator light detection,we propose an improved You Only Look Once vision 5(YOLOv5)power plant indicator light detection algorithm.The algorithm improves the feature extraction ability based on YOLOv5s.First,our algorithm enhances the ability of the network to perceive small objects by combining attention modules for multi-scale feature extraction.Second,we adjust the loss function to ensure the stability of the object frame during the regression process and improve the conver-gence accuracy.Finally,transfer learning is used to augment the dataset to improve the robustness of the algorithm.The experimental results show that the average accuracy of the proposed squeeze-and-excitation YOLOv5s(SE-YOLOv5s)algorithm is increased by 4.39%to 95.31%compared with the YOLOv5s algorithm.The proposed algorithm can better meet the engineering needs of power plant indicator light detection.

关键词： You Only Look Once vision 5(YOLOv5) attention module loss function transfer learning object detection system

Leaky Cable Fixture Detection in Railway Tunnel Based on RW DCGAN and Compressed GS-YOLOv5

在线全文

学校读者我要写书评

暂无评论

Intelligent Automation & Soft Computing 2023年第7期37卷 1163-1180页

作者： Suhang Li yunzuo zhang Ruixue Liu Jiayu zhang Zhouchen Song Yutai Wang School of Information Science and Technology Shijiazhuang Tiedao UniversityShijiazhuang050043China Australian National University Canberra2600ACTAustralia

The communication system of high-speed trains in railway tunnels needs to be built with leaky cables fixed on the tunnel wall with special fixtures.To ensure safety,checking the regular leaky cable fixture is necessary to elimi-nate the potential danger.At present,the existing fixture detection algorithms are difficult to take into account detection accuracy and speed at the same time.The faulty fixture is also insufficient and difficult to obtain,seriously affecting the model detection effect.To solve these problems,an innovative detection method is proposed in this paper.Firstly,we presented the Res-Net and Wasserstein-Deep Convolution GAN(RW-DCGAN)to implement data augmentation,which can enable the faulty fixture to export more high-quality and irregular images.Secondly,we proposed the Ghost SENet-YOLOv5(GS-YOLOv5)to enhance the expression of fixture feature,and further improve the detection accuracy and speed.Finally,we adopted the model compression strategy to prune redundant channels,and visualized training details with Grad-CAM to verify the reliability of our model.Experimental results show that the algorithm model is 69.06%smaller than the original YOLOv5 model,with 70.07%fewer parameters,2.1%higher accuracy and 14.82 fps faster speed,meeting the needs of tunnel fixture detection.

关键词： Leaky cable fixture detection RW-DCGAN compressed strategy GS-YOLOv5

A Review of Research on Person Re-identification in Surveillance Video

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Mechanical Engineering Science 2023年第2期5卷 1-7页

作者： yunzuo zhang Weiqi LIAN School of Information science and Technology Shijiazhuang Tiedao UniversityShijiazhuangHebei050043China

Person re-identification has emerged as a hotspot for computer vision research due to the growing demands of social public safety requirements and the quick development of intelligent surveillance networks.Person re-identification(Re-ID)in video surveillance system can track and identify suspicious people,track and statistically analyze persons.The purpose of person re-identification is to recognize the same person in different cameras.Deep learning-based person re-identification research has produced numerous remarkable outcomes as a result of deep learning's growing popularity.The purpose of this paperis to help researchers better understand where person re-identification research is at the moment and where it is headed.Firstly,this paper arranges the widely used datasets and assessment criteria in person re-identification and reviews the pertinent research on deep learning-based person re-identification techniques conducted in the last several years.Then,the commonly used method techniques are also discussed from four aspects:appearance features,metric learning,local features,and adversarial learning.Finally,future research directions in the field of person re-identification are outlooked.

关键词： Person re-identification Deep learning Metric learning Local features Adversarial learning

A Review of Lane Detection Based on Deep Learning Methods

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Mechanical Engineering Science 2023年第2期5卷 37-48页

作者： yunzuo zhang Zhiwei TU Fenfen LYU School of Information science and Technology Shijiazhuang Tiedao UniversityShijiazhuangHebei050043China Sports Technical School of Hebei Sports Bureau ShijiazhuangHebei050041China

Lane detection is animportant aspect of autonomous driving,aiming to ensure that vehicles accurately understand road structures as well as improve their ability to drive in complex traffic environments.In recent years,lane detection tasks based on deep learning methods have made significant progress in detection accuracy.In this paper,we provide a comprehensive review of deep learning-based lane detection tasks in recent years.First,we introduce the background of the lane detection task,including lane detection,the lane datasets and the factors affecting lane detection.Second,we review the traditional and deep learning methods for lane detection,and analyze their features in detail while classifying the different methods.In the deep learning methods classification section,we explore five main categories,including segmentation-based,object detection,parametric curves,end-to-end,and keypoint-based methods.Then,some typical models are briefly compared and analyzed.Finally,in this paper,based on the comprehensive consideration of current lane detection methods,we put forward the current problems still faced,such as model generalization and computational cost.At the same time,possible future research directions are given for extreme scenarios,model generalization and other issues.

关键词： Deep learning Lane detection Image segmentation Object detection Parametric curves