检索结果-内蒙古大学图书馆

IEEE Transactions on Biometrics, Behavior, and Identity Science 2021年第1期3卷 31-43页

作者： Doukas, Michail Christos Koujan, Mohammad Rami Sharmanska, Viktoriia Roussos, Anastasios Zafeiriou, Stefanos Department of Computing Imperial College London LondonN1C 4AG United Kingdom College of Engineering Mathematics and Physical Sciences University of Exeter ExeterEX4 4PY United Kingdom Department of Computing Imperial College London LondonSW7 2BU United Kingdom Institute of Computer Science Heraklion700 13 Greece

Facial video re-targeting is a challenging problem aiming to modify the facial attributes of a target subject in a seamless manner by a driving monocular sequence. We leverage the 3D geometry of faces and Generative Adversarial Networks (GANs) to design a novel deep learning architecture for the task of facial and head reenactment. Our method is different to purely 3D model-based approaches, or recent image-based methods that use deep Convolutional Neural Networks (DCNNs) to generate individual frames. We manage to capture the complex non-rigid facial motion from the driving monocular performances and synthesise temporally consistent videos, with the aid of a sequential Generator and an ad-hoc Dynamics Discriminator network. We conduct a comprehensive set of quantitative and qualitative tests and demonstrate experimentally that our proposed method can successfully transfer facial expressions, head pose and eye gaze from a source video to a target subject, in a photo-realistic and faithful fashion, better than other state-of-the-art methods. Most importantly, our system performs end-to-end reenactment in nearly real-time speed (18 fps). © 2019 IEEE.

关键词： Generative adversarial networks

来源：评论

学校读者我要写书评

暂无评论

Six Convolutional Layered deep Convolutional Neural Network Based real time Single image Super Resolution

Six Convolutional Layered Deep Convolutional Neural Network ...

引用

Smart Electronics and Communication (ICOSEC), International Conference on

作者： M. Shyamala Devi J. Arun Pandian Rahul Kumar Thakur Vinod Babu Vemuri Javvaji Naga Bharath Computer Science & Engineering Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and Technology Chennai Tamilnadu India School of Information Technology and Engineering Vellore Institute of Technology Vellore India

The super-resolving images investigation has advanced in recent years through the use of cutting-edge deep learning-based architectures. Numerous previously documented super-resolution-based solutions need the most advanced and top-tier Graphics processing Units (GPUs) to execute picture super-resolution. With the growing advancement of technology, this research focuses on suggesting the needed quantity of convolutional layers for creating the real time super resolution of single image using Convolutional Neural Network. The proposed Six Convolutional Layered deep Convolutional Neural Network (6CL-DCNN) to predict the super resolution of images with high accuracy and ideal Peak Signal-to-Noise Ratio. The dataset extracted from ***(http://***/Research/Projects/CS/vision/grouping/BSR/BSR_***). The dataset contains 800 images with the combination of low resolution and high-resolution images having the image resolution of 300 * 300 pixels. The proposed 6CL-DCNN built with single input layer followed by six convolutional layers followed by single lambda optimized output layer that predicts the super resolution of both high resolution images and low resolution images. Python was used through 500 training iterations and a 64-bit block size on a NVidia Geforce Tesla V100 GPU workstation. The processed low resolution images and high-resolution images are applied with proposed 6CL-DCNN model and also the performance is compared using Peak Signal to Noise Ratio with other optimized output layers. Experimental results shows that the proposed model 6CL-DCNN shows the maximum Peak Signal to Noise Ratio of 48 dB when compared to other optimized output layers.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Incorporating Mtcnn and deepface: A Novel Hybrid deep learning Method for Human Face Detection and Recognition

SSRN

引用

SSRN 2023年

作者： Honey Singh, Sukhwinder Punjabi University India Guru Hargobind Sahib Khalsa Girls College India

The realm of face detection has become a focal point of extensive research, driven by its diverse applications spanning computer vision, communication, and automatic control systems. realizing real-time recognition of multiple faces within embedded systems poses a formidable challenge due to the intricate computational demands involved. This challenge necessitates a deep exploration of facets such as face detection, expression recognition, face tracking, and pose estimation. Accurately identifying a face from a single image stands as the core challenge, primarily due to the non-rigid nature of faces, resulting in variations in size, shape, color, and more. Furthermore, the complexity of face detection amplifies when confronted with unclear images, occlusions, suboptimal lighting conditions, off-angle poses, and various other factors. This study presents an innovative framework for multiple face recognition. Through extensive experiments, the system's prowess in simultaneously recognizing up to 10 different human face poses in real time was showcased, achieving remarkable processing speeds as low as 0.21 seconds. The system demonstrated an impressive minimum recognition rate of 93.15%, underscoring the effectiveness of the proposed methodology. While the primary emphasis lies on frontal human faces, the system is adept at handling poses beyond the frontal orientation, marking a significant advancement in the domain of face detection and recognition. © 2023, The Authors. All rights reserved.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

A deep learning Module Design for Workspace Identification in Manufacturing Industry 3

A Deep Learning Module Design for Workspace Identification i...

引用

3rd International Conference on Artificial Intelligence in Information and Communication (IEEE ICAIIC)

作者： Kim, Jeong-Su Lee, Dong Myung Tongmyong Univ Grad Sch Dept Comp & Media Engn Busan South Korea Tongmyong Univ Dept Comp Engn Busan South Korea

ISBN: (纸本)9781728176383

In this paper, in order to solve various problems occurring in the workspace, a deep learning-based workspace identification module was designed, and the performance was analyzed through an experiment on the recognition accuracy according to the configuration of the training dataset and the number of training. The data model of the designed deep learning module is ResNet18, and after setting up three dataset strategies, a dataset using five types of workspaces of the manufacturing industry was selected. In terms of the average top 5 and all training, strategy 2 was 812% and 76.4%, respectivel, confirming that it was the best among the 3 strategies. In the future, after upgrading the designed module, it is planned to implement a module with real-time workspace identification performance level of practical use in a mobile environment with an image input device installed.

关键词： scene recognition deep learning convolutional neural network datasets places365 manufacturing workspace

来源：评论

学校读者我要写书评

暂无评论

real-time defect detection network for polarizer based on deep learning

引用

JOURNAL OF INTELLIGENT MANUFACTURING 2020年第8期31卷 1813-1823页

作者： Liu, Ruizhen Sun, Zhiyi Wang, Anhong Yang, Kai Wang, Yin Sun, Qianlai Taiyuan Univ Sci & Technol Sch Mat Sci & Engn Taiyuan 030024 Shanxi Peoples R China Taiyuan Univ Sci & Technol Sch Elect Informat Engn Taiyuan 030024 Shanxi Peoples R China

Quality analysis of the polarizer of a production line can be performed using image processing technology. The existing method of detecting defective images based on deep learning can ensure accurate classification;however, its detection speed is low, the model requires a large amount of memory, and it is difficult to meet the real-time requirements of online detection systems when hardware resources are limited. Therefore, in this study a lightweight polarizer defect detection network, called DDN, was developed based on deep learning. First, a parallel module was designed to build the network. This module has two main advantages. First, it mixes different convolution template sizes, and can fuse the features of different scales and extract more defect features than the traditional convolution layer. Second, depthwise separable convolution is used to replace full convolution in this module, which significantly reduces the number of parameters and the multiply-accumulate operations. Finally, a global average pooling (GAP) layer is used instead of a fully connected layer. The GAP layer has no parameters to optimize, which substantially reduces the number of network parameters. Experimental results show that the proposed method is better than existing methods in terms of classification speed, precision, and memory consumption for polarizer detection, and can satisfy real-time requirements.

关键词： Automatic testing image classification Defect detection Global average pooling Parallel module

来源：评论

学校读者我要写书评

暂无评论

Layer-Wise Multi-Defect Detection for Laser Powder Bed Fusion Using deep learning Algorithm with Visual Explanation

SSRN

引用

SSRN 2023年

作者： Zhao, Yingjian Ren, Hang Zhang, Yuhui Wang, Chengyun Long, Yu Institute of Laser Intelligent Manufacturing and Precision Processing School of Mechanical Engineering Guangxi University Guangxi Nanning530004 China

In Laser Powder Bed Fusion (LPBF), it is a major challenge to obtain detailed spatial information on different powder bed defects in real-time and simultaneously. deep learning (DL) algorithms under the field of Machine learning (ML) have promoted the intelligent development of the powder bed defect detection method. However, they still need to be further evaluated in terms of detection accuracy and time delay, training data overhead, and model robustness under complex environments. Also, the DL model usually treated as a black-box demands further explanation. Herein, three advanced DL models are constructed using bounding boxes to locate multi-defects quickly and accurately for the powder spreading of LPBF. High detection accuracy is achieved with limited training samples through both data augmentation aiming at expanding image samples and model-based Transfer learning (TL) used for transferring from the source domain to the target domain by reusing trained models. Further, the data augmentation method is employed to generate low-resolution images with interference to test the robustness of the detection model in harsh environments. Besides, visual feature maps and saliency maps were generated with the Detector Randomized Input Sampling for Explanation (D-RISE) method to help understand the validity of the defect detection process of the DL model. Overall, this work shows that the proposed multi-defect detection algorithms can provide comprehensive information on powder bed defects accurately and quickly. © 2023, The Authors. All rights reserved.

关键词： Process monitoring

来源：评论

学校读者我要写书评

暂无评论

FaceMask: A New image Dataset for the Automated Identification of People Wearing Masks in the Wild

引用

SENSORS 2022年第3期22卷 896-896页

作者： Vrigkas, Michalis Kourfalidou, Evangelia-Andriana Plissiti, Marina E. Nikou, Christophoros Univ Western Macedonia Dept Commun & Digital Media Kastoria 52100 Greece Univ Ioannina Dept Comp Sci & Engn Ioannina 45110 Greece

The rapid spread of the COVID-19 pandemic, in early 2020, has radically changed the lives of people. In our daily routine, the use of a face (surgical) mask is necessary, especially in public places, to prevent the spread of this disease. Furthermore, in crowded indoor areas, the automated recognition of people wearing a mask is a requisite for the assurance of public health. In this direction, image processing techniques, in combination with deep learning, provide effective ways to deal with this problem. However, it is a common phenomenon that well-established datasets containing images of people wearing masks are not publicly available. To overcome this obstacle and to assist the research progress in this field, we present a publicly available annotated image database containing images of people with and without a mask on their faces, in different environments and situations. Moreover, we tested the performance of deep learning detectors in images and videos on this dataset. The training and the evaluation were performed on different versions of the YOLO network using Darknet, which is a state-of-the-art real-time object detection system. Finally, different experiments and evaluations were carried out for each version of YOLO, and the results for each detector are presented.

关键词： face-mask mask detector dataset neural networks Darknet YOLO

来源：评论

学校读者我要写书评

暂无评论

Hierarchical Residual Embedding Network for real-time De-focus Blur Detection 2

Hierarchical Residual Embedding Network for Real-Time De-foc...

引用

2020 2nd International Conference on Computer, Communications and Mechatronics Engineering, CCME 2020

作者： Zhang, Chunlei Nie, Jing Zheng, Xiao Electronic Engineering College Naval University of Engineering Wuhan430033 China Teaching and Research Support Center Naval University of Engineering Wuhan430033 China School of Computer Science National University of Defense Technology Changsha410073 China

Defocus blur detection, as an important pre-processing step of image processing, has attracted more and more attention. Albeit great success has been made, there are still several challenges for accurate defocus blur detection, such as the interference of background clutter, sensitivity to scales, missing boundary details and large computational burden. For handling these issues, we present a deep neural network which hierarchically embeds residual learning blocks for defocus blur detection. Based on the feature pyramid structure, we extract deep features with varying scales via utilizing a backbone fully convolutional network and generate a coarse score map by using the last layer of feature maps. Then we design a hierarchical residual embedding module to fuse different levels of features in a layer-wise manner. By embedding different layer-wise features in the top-down pathway, coarse-level semantic information from the deep layers can be seamlessly propagated to shallow layers, while fine details in the shallow layers can be used to refine the boundary between out-of-focus and in-focus regions. For each layer, a side output is generated by using a residual learning block. For capturing multi-scale information, the multiple side outputs of different layers are fed into a designed fusion block for yielding the final blur map result. Experimental results on two commonly used datasets show that our proposed network can more accurately locate the defocus blur regions with sharpened details being well preserved when compared to other previous state-of-the-arts. In addition, our approach is fast as well and can run at a speed of more than 25 FPS when processing an image with size 427 x 640. © Published under licence by IOP Publishing Ltd.

关键词： deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Improved Classification Approach for Fruits and Vegetables Freshness Based on deep learning

引用

SENSORS 2022年第21期22卷 8192页

作者： Mukhiddinov, Mukhriddin Muminov, Azamjon Cho, Jinsoo Gachon Univ Dept Comp Engn Seongnam 13120 South Korea

Classification of fruit and vegetable freshness plays an essential role in the food industry. Freshness is a fundamental measure of fruit and vegetable quality that directly affects the physical health and purchasing motivation of consumers. In addition, it is a significant determinant of market price;thus, it is imperative to study the freshness of fruits and vegetables. Owing to similarities in color, texture, and external environmental changes, such as shadows, lighting, and complex backgrounds, the automatic recognition and classification of fruits and vegetables using machine vision is challenging. This study presents a deep-learning system for multiclass fruit and vegetable categorization based on an improved YOLOv4 model that first recognizes the object type in an image before classifying it into one of two categories: fresh or rotten. The proposed system involves the development of an optimized YOLOv4 model, creating an image dataset of fruits and vegetables, data argumentation, and performance evaluation. Furthermore, the backbone of the proposed model was enhanced using the Mish activation function for more precise and rapid detection. Compared with the previous YOLO series, a complete experimental evaluation of the proposed method can obtain a higher average precision than the original YOLOv4 and YOLOv3 with 50.4%, 49.3%, and 41.7%, respectively. The proposed system has outstanding prospects for the construction of an autonomous and real-time fruit and vegetable classification system for the food industry and marketplaces and can also help visually impaired people to choose fresh food and avoid food poisoning.

关键词： fruit classification fruit and vegetable freshness YOLOv4 computer vision object detection deep learning convolutional neural network

来源：评论

学校读者我要写书评

暂无评论

Statistical Analysis of Medium-Scale Traveling Ionospheric Disturbances Over Japan Based on deep learning Instance Segmentation

引用

SPACE WEATHER-THE INTERNATIONAL JOURNAL OF RESEARCH AND APPLICATIONS 2022年第7期20卷 e2022SW003151-e2022SW003151页

作者： Liu, Peng Yokoyama, Tatsuhiro Fu, Weizheng Yamamoto, Mamoru Kyoto Univ Res Inst Sustainable Humanosphere Uji Japan

Medium-scale traveling ionospheric disturbances (MSTIDs) are observed as parallelly arrayed wavelike perturbations of Total Electron Content (TEC) in ionospheric F region leading to satellite navigation error and communication signal scintillation. The observation method for MSTIDs, detrended TEC (dTEC) map, summarizes the perturbation component of TEC having the merits of full-time and two-dimensional. However, previous automatic processing methods for dTEC map cannot discriminate MSTIDs from other irregular ionospheric perturbations intelligently. With the development of artificial intelligence in recent years, deep learning approach is expecting to clarify the controversy of MSTID external dependence (season and solar/geomagnetic activity) under debating for decades. Therefore, this research proposes a real-time processing algorithm for dTEC maps based on Mask Region-Convolutional Neural Network (R-CNN) model of deep learning instance segmentation to detect wavelike perturbations intelligently with an accuracy of about 80% and a processing speed of about 8 fps. Then isolated perturbations are eliminated and only MSTID waveforms are chosen to obtain statistical characteristics of MSTIDs. With this algorithm, we analyzed up to 1,209,600 dTEC maps from 1997 to 2019 over Japan automatically and established a database of hourly averaged MSTID characteristics. This research introduces the partial correlation coefficient for the first time to clarify the solar/geomagnetic activity dependence of MSTID characteristics which is independent with each other.

关键词： MSTID ionospheric irregularity wavelike perturbation statistical analysis deep learning instance segmentation

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：