检索结果-内蒙古大学图书馆

10th IEEE International Conference on Computer Science and Network Technology, ICCSNT 2022

作者： Liang, Ranran Ning, Tao Jia, Xiangkun Institute of Computer Science Dalian Minzu University State Ethnic Affairs Commission Key Laboratory of Big Data Applied Technology Dalian China

ISBN: (数字)9781665470612

ISBN: (纸本)9781665470612

In order to decrease in the accuracy of traffic sign recognition due to dim light at night, a novel YOLOv5 algorithm is proposed in this paper, which adds the Improved Adaptive Histogram Equalization (IAHE) method to image pre-processing, adjusts image brightness and contrast to highlight the related information about traffic signs. In view of the higher requirements of driving assistance system on the processing speed of recognition model and the standard convolution mode of backbone network is improved to the depth separable convolution method for model lightweight processing. To address the problem of wrong and missed detection generated by prediction box suppression, the conventional NMS is replaced by Weighted Boxes Fusion WBF to generate prediction box. Experiments have demonstrated that the improved algorithm has improved detection accuracy and decreased processing time for a single image compared with the original YOLOv5 algorithm in the self-built night environment dataset. © 2022 IEEE.

关键词： Convolution

来源：评论

学校读者我要写书评

暂无评论

Countries flags detection based on local context network and color features

引用

MULtimeDIA TOOLS AND APPLICATIONS 2021年第10期80卷 14753-14765页

作者： Said, Yahia Barr, Mohammad Northern Border Univ Coll Engn Elect Engn Dept Ar Ar Saudi Arabia Univ Monastir Fac Sci Monastir Lab Elect & Microelect LR99ES30 Monastir Tunisia

Countries flags are characterized by a combination of special colors. Building an automatic country flag detector is a hard task because of many challenges like deformation and difference in point of view. Motivated by the unique feature of the country flag colors and the power of deep learning models, we propose to use color-based features and a Convolutional Neural Network (CNN) with a special local context neural network to perform the countries flags detection task. The proposed approach aims to enhance the performance of the ordinary Convolutional Neural Network by adding a local context neural network to enhance the localization task and adding a color-based descriptor to enhance the identification task. The color-based descriptor was used to focus on the color features because of its importance for the studied task. The Convolutional Neural Network was proposed to extract more relevant features for both localization and identification tasks. The local context network was used to localize the flag in the image. In order to train and evaluate the proposed approach, we propose to build a custom dataset for the world countries' flags. The proposed dataset counts 100 images for each country flag with a total of 20,000 images. The evaluation of the proposed approach proves its efficiency by achieving a mean Average Precision of 89.5% and a real-time processing speed. The achieved results have proved the efficiency of the proposed method. The proposed enhancement was very effective that allows the achievement of high accuracy.

关键词： Flag detection deep learning Convolutional Neural Network Local context network Color features

来源：评论

学校读者我要写书评

暂无评论

deep-learning real-time phase retrieval of imperfect diffraction patterns from X-ray free-electron lasers

arXiv

引用

arXiv 2024年

作者： Lee, Sung Yun Cho, Do Hyung Jung, Chulho Sung, Daeho Nam, Daewoong Kim, Sangsoo Song, Changyong Department of Physics POSTECH Pohang37673 Korea Republic of Photon Science Center POSTECH Pohang37673 Korea Republic of Center for Ultrafast Science on Quantum Matter Max Planck POSTECH/Korea Research Initiative Pohang37673 Korea Republic of Pohang Accelerator Laboratory POSTECH Pohang37673 Korea Republic of

Machine learning is attracting surging interest across nearly all scientific areas by enabling the analysis of large datasets and the extraction of scientific information from incomplete data. Data-driven science is rapidly growing, especially in X-ray methodologies, where advanced light sources and detection technologies accumulate vast amounts of data that exceed meticulous human inspection capabilities. Despite the increasing demands, the full application of machine learning has been hindered by the need for data-specific optimizations. In this study, we introduce a new deep-learning-based phase retrieval method for imperfect diffraction data. This method provides robust phase retrieval for simulated data and performs well on weak-signal single-pulse diffraction data from X-ray free-electron lasers. Moreover, the method significantly reduces data processing time, facilitating real-time image reconstructions that are crucial for high-repetition-rate data acquisition. Thus, this approach offers a reliable solution to the phase problem and is expected to be widely adopted across various research *** Codes 68T07 © 2024, CC BY-NC-ND.

关键词： Pulse repetition rate

来源：评论

学校读者我要写书评

暂无评论

Enhanced Accident Detection in Low-Light Conditions Through Analysis of CCTV Footage in real-time 7th

Enhanced Accident Detection in Low-Light Conditions Through ...

引用

7th International Conference on Soft Computing and Signal processing, ICSCSP 2024

作者： Goutham Kumar Reddy, S. Sampreeth, M. Sai Vardhan Reddy, K. Balamurugan, P. Mummoorthy, A. SRM Institute of Science and Technology KTR Chennai India Malla Reddy College of Engineering and Technology Hyderabad India

ISBN: (纸本)9789819609239

In today’s interconnected world, ensuring road safety is of paramount importance, especially in challenging conditions such as low-light environments where accidents are more likely to occur. This paper introduces an innovative approach to elevate accident detection capabilities in such conditions through the use of advanced computer vision and machine learning methodologies. The proposed system integrates state-of-the-art image processing algorithms and the YOLO (You Only Look Once) algorithm, renowned for its real-time object detection capabilities. By analyzing CCTV footage in real-time, the system enhances visibility and accuracy in identifying incidents, contributing significantly to proactive incident management and emergency response times. Through rigorous testing and evaluation, the system achieved an impressive accuracy rate of 94.06%, demonstrating its effectiveness and reliability in real-world scenarios. The integration of a web application-based alert mechanism further enhances the system's capabilities by enabling rapid notification of relevant authorities and nearby individuals in case of accidents. Overall, this paper represents a significant leap forward in road safety technologies, showcasing the transformative potential of cutting-edge technologies in addressing critical challenges in urban safety and surveillance. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： deep learning

来源：评论

学校读者我要写书评

暂无评论

Enabling microservices management for deep learning applications across the Edge-Cloud Continuum 33

Enabling microservices management for Deep Learning applicat...

引用

33rd IEEE International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD)

作者： Houmani, Zeina Balouek-Thomert, Daniel Caron, Eddy Parashar, Manish Univ Lyon Inria Avalon Team LIP Lab UMR CNRS ENS Lyon Lyon France Univ Utah Sci Comp Imaging SCI Inst Salt Lake City UT 84112 USA

ISBN: (纸本)9781665443012

deep learning has shifted the focus of traditional batch workflows to data-driven feature engineering on streaming data. In particular, the execution of deep learning workflows presents expectations of near-real-time results with user-defined acceptable accuracy. Meeting the objectives of such applications across heterogeneous resources located at the edge of the network, the core, and in-between requires managing trade-offs between the accuracy and the urgency of the results. However, current data analysis rarely manages the entire deep learning pipeline along the data path, making it complex for developers to implement strategies in real-world deployments. Driven by an object detection use case, this paper presents an architecture for time-critical deep learning workflows by providing a data-driven scheduling approach to distribute the pipeline across Edge to Cloud resources. Furthermore, it adopts a data management strategy that reduces the resolution of incoming data when potential trade-off optimizations are available. We illustrate the system's viability through a performance evaluation of the object detection use case on the Grid'5000 testbed. We demonstrate that in a multi-user scenario, with a standard frame rate of 25 frames per second, the system speed-up data analysis up to 54.4% compared to a Cloud-only-based scenario with an analysis accuracy higher than a fixed threshold.

关键词： Cloud computing Edge computing Microservices Task allocation real-time processing Computing Continuum deep learning

来源：评论

学校读者我要写书评

暂无评论

deep-learning Based Three-Dimensional Analysis of Bubble Flows From Light Field images 29

Deep-Learning Based Three-Dimensional Analysis of Bubble Flo...

引用

2022 29th International Conference on Nuclear Engineering, ICONE 2022

作者： Zhang, Heng Lyu, Xue Hang, Qin Du, Yulu Liu, Bo Liu, Qun Wang, Guoyin Liu, Jinchao College of Computer Science and Technology Chongqing University of Posts and Telecommunications Chongqing China Simulation Technology Co. Ltd Guangdong Shenzhen China

ISBN: (纸本)9780791886502

An accurate measurement for the modality and motion parameters of bubbles is of great significance. In the existing bubble measurement methods, there are some problems that are desirable to be solved, such as system complexity, poor real-time performance, and the deviation due to two-dimensional measurement. To overcome these challenges, we developed a novel three-dimensional analysis method based on light field imaging diagnosis and deep learning algorithm. Different from traditional two-dimensional reconstruction, the bubble depth can be computed from light field images directly through digital refocusing technology. After calibrating, the conversion between the camera coordinate system and the real-world coordinate system is achieved through the sharpness evaluation algorithm. According to the corresponding relationship, the refocused image could be calibrated to the actual position in real world. Combined with the Multi-input Residual Convolution Neural Network, (MRCNN), the bubble depth could be computed fully automated from given images at high accuracy. Based on the above works, the three-dimensional reconstruction model for bubble flow can be established by coupling the depth and the parameters extracted through bounding boxes. The proposed method solves the problem that conventional imaging can only perform the two-dimensional measurement, which contributes to the error during the measurement process. Results show a promising performance on the three-dimensional reconstruction of bubble flow, validating the feasibility of the three-dimensional measurement method for bubbles in gas-liquid two-phase flow based on light field imaging diagnosis and deep learning algorithm. © 2022 by ASME.

关键词： learning algorithms

来源：评论

学校读者我要写书评

暂无评论

Dynamic Task Division and Allocation in Mobile Edge Computing Systems: A Latency Oriented Approach via deep Q-learning Network 23

Dynamic Task Division and Allocation in Mobile Edge Computin...

引用

IEEE 23rd International Conference on High Performance Switching and Routing (IEEE HPSR)

作者： Tan, Pengcheng Li, Yang Dai, Minghui Wu, Yuan Univ Macau State Key Lab Internet Things Smart City Taipa Macao Peoples R China Zhuhai UM Sci & Technol Res Inst Zhuhai Peoples R China

ISBN: (纸本)9781665406079

With the rapid development of Internet of Things (IoTs), various sensors are deployed to collect different physical information. Smart surveillance is one of applications by analyzing the real-time video generated by camera sensors. However, due to the limited computing capability of camera sensors, running video analysis models (e.g., AlexNet and YOLO3) on camera sensors directly consumes a lot of computing time. In addition, transferring video to the remote cloud suffers a long-distance transmission latency. Fortunately, edge computing has been considered as a promising solution for enabling computation-intensive yet latency-sensitive applications at resource-constrained devices. Thanks to edge computing, camera sensors can upload video to different edge servers employed at the edge of networks for processing. Moreover, the lightweight Kubernetes for edge computing, i.e., K3s, enable a fine-grained task division and parallel computing. In this paper, we consider a heterogeneous edge cooperative video analysis, i.e., face recognition, with the objective of minimizing the processing latency. Specifically, we use a deep Q-learning network (DQN) to dynamically adjust the size of pieces video allocated to different edge servers connected via wireless networks. In addition, to improve the resource utilization of edge servers and reduce the processing latency, each edge server further divides the received video into multiple segments that are processed by different containers in parallel. To validate the effectiveness of our scheme, we implement a small-scale prototype system and conduct numerous experiments. Experimental results show that our proposed algorithm outperforms the other four schedule schemes by testing on the tasks of face recognition and pose recognition.

关键词： Computing resources Task execution time Edge intelligence Task division deep learning

来源：评论

学校读者我要写书评

暂无评论

deep Universal Blind image Denoising 25

Deep Universal Blind Image Denoising

引用

25th International Conference on Pattern Recognition (ICPR)

作者： Soh, Jae Woong Cho, Nam Ik Seoul Natl Univ INMC Dept ECE Seoul South Korea

ISBN: (纸本)9781728188089

image denoising is an essential part of many image processing and computer vision tasks due to inevitable noise corruption during image acquisition. Traditionally, many researchers have investigated image priors for the denoising, within the Bayesian perspective based on image properties and statistics. Recently, deep convolutional neural networks (CNNs) have shown great success in image denoising by incorporating large-scale synthetic datasets. However, they both have pros and cons. While the deep CNNs are powerful for removing the noise with known statistics, they tend to lack flexibility and practicality for the blind and real-world noise. Moreover, they cannot easily employ explicit priors. On the other hand, traditional nonlearning methods can involve explicit image priors, but they require considerable computation time and cannot exploit large-scale external datasets. In this paper, we present a CNN-based method that leverages the advantages of both methods based on the Bayesian perspective. Concretely, we divide the blind i mage denoising problem into sub-problems and conquer each inference problem separately. As the CNN is a powerful tool for inference, our method is rooted in CNNs and propose a novel design of network for efficient inference. With our proposed method, we can successfully remove blind and real-world noise, with a moderate number of parameters of universal CNN.

关键词： image denoising

来源：评论

学校读者我要写书评

暂无评论

A Performance Study on deep learning Covid-19 Prediction through Chest X-Ray image with ResNet50 Model 2

A Performance Study on Deep Learning Covid-19 Prediction thr...

引用

2nd International Conference on Advances in Electrical, Computing, Communication and Sustainable Technologies, ICAECT 2022

作者： Kumar, Dhirendra Sharma, Pulkit Anupama, Anupama Sharma, Parth Delhi Technological University Department of Applied Mathematics Delhi India Delhi Technological University Department of Electrical Engineering Delhi India

ISBN: (纸本)9781665411202

The COVID-19 epidemic has claimed many lives throughout the world and constitutes an unprecedented public health concern. The key challenge in early detection of the corona virus is early detection. And the main obstacle was the similarity of COVID-19 symptoms to flu symptoms. With the goal of saving human lives and stemming the spread of a worldwide pandemic, an accurate and speedy analysis of COVID-19-induced pneumonia has now taken centre stage. Responding this urgent concern and to reduce the burden as well as chances of faulty manual diagnosis, several deep learning approaches are developed to conduct early diagnosis. Based on the availability of reliable patient's records, an accepted technique is pre-trained deep learning prediction approach through patient's chest X-Rays. Convenience of this approach led development of a number of novel deep learning-based lung screening technologies. However, little emphasis is placed on ensuring the quality of their output. Pre-trained deep learning systems will be used in this project to evaluate their ability to recognise and diagnose disorders. To categorise COVID and normal pictures, a neural network-based ResNet50 architecture is presented. The implementation is based on the normal, COVID, and lung opacity datasets. For data pre-processing, imageDataGenerator is used, which rescales, flips, and modifies the range to meet the model. To categorise the x- ray images, the suggested method ResNet50 architecture is used. Performance matrices like precision, accuracy, recall, as well as F1-score are examined to verify the algorithm's usefulness. The suggested technique has an accuracy of 80%, indicating that the proposed model is quite good in classifying COVID and normal x-ray pictures. This research will have a significant influence on real-time since it will accurately diagnose COVID in less time, perhaps lowering the mortality rate. © 2022 IEEE.

关键词： COVID-19

来源：评论

学校读者我要写书评

暂无评论

Pest detection and classification to reduce pesticide use in fruit crops based on deep neural networks and image processing 19

Pest detection and classification to reduce pesticide use in...

引用

19th Workshop on Information processing and Control (RPIC)

作者： Suarez, Agustina Soledad Molina, Romina Ramponi, Giovanni Petrino, Ricardo Bollati, Luciana Sequeiros, Daniel Univ Nacl San Luis LEIS Dept Elect Av Ejercito Andes 950D5700 San Luis Argentina Univ Trieste IPL Dipartimento Ingn & Architettura Piazzale Europa 1 I-34127 Trieste TS Italy Nectras Av Belgrano 758 Sunchales Santa Fe Argentina

ISBN: (纸本)9781665414364

The aim for organic farming is obtaining food of the highest quality, avoiding synthetic chemicals, protecting the environment and preserving the fertility of the land. In this context, effective pest control allows to reduce yield loss and pesticides application producing pollution-free vegetables. In fruit crops, Carpocapsa is the main pest present in pear, apple, walnut and quince trees. This insect produces irreversible damage to the fruit, since the larvae feed the seeds inside the fruit. In this paper, we present automatic pest detection and classification in the context of fruit crops based on image processing and deep Neural Networks, employing an image collection obtained from in-field traps. Due to the limited size of the data set, we perform data augmentation to increase the number of images for training, to prevent over-fitting and to improve the deep neural network learning rate. Results showed an overall accuracy of 94.8%, while precision and recall scores for the class related with the moth were around 97.2% and 93.6% respectively, demonstrating the efficacy of this type of classifier proposed for pest detection. An inference time of 40 ms per image for the deep neural network classifier has been reached.

关键词： image processing deep neural networks CNN classification pest detection

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：