检索结果-内蒙古大学图书馆

8th International Conference on Information System Design and Intelligent applications, ISDIA 2024

作者： Nahar, Shreya Jain, Tanishka R. Shah, Mihir Dilip, Golda Department of Computer Science and Engineering SRMIST Vadapalani Chennai India

ISBN: (纸本)9789819748914

The automatic description of the content of an image is a challenge that combines computer vision with natural language processing in the field of artificial intelligence. Most existing models have limitations in accurately identifying subtle elements of an image and counting the number of people present. To address this challenge, we introduce the natural-language scene description network (NSDN), which incorporates recent developments in computer vision and machine translation to generate descriptive phrases that effectively depict images. This technology has the potential to improve the efficiency, capacity, reliability, and safety of crowd management tasks, particularly in diverse and adaptive crowd situations. Despite challenges such as clutter, occlusion, non-uniform object scale, and irregular object distribution, YOLO shows promise for intelligent crowd counting and analysis in images. This article reviews, categorizes, analyzes distinguishing features, and extensively assesses the effectiveness of crowd-counting methods that rely on convolutional neural networks and provides a detailed analysis of their performance. © The Author(s).

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Data Augmentation-based Novel Deep Learning Method for Deepfaked images Detection

引用

ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND applications 2024年第11期20卷 1-15页

作者： Iqbal, Farkhund Abbasi, Ahmed Javed, Abdul rehman Almadhor, Ahmad Jalil, Zunera Anwar, Sajid Rida, Imad Zayed Univ Coll Technol Innovat Abu Dhabi 144534 U Arab Emirates Air Univ Dept Creat Technol PAF Complex Islamabad 56300 Pakistan Lebanese Amer Univ Dept Elect & Comp Engn Byblos Lebanon Jouf Univ Coll Comp & Informat Sci Dept Comp Engn & Networks Sakaka 42421 Saudi Arabia Air Univ Dept Cyber Secur PAF Complex Islamabad 56300 Pakistan Inst Management Sci Peshawar Peshawar 25130 Pakistan Univ Technol Compiegne BMBI Lab F-60203 Compiegne France

Recent advances in artificial intelligence have led to deepfake images, enabling users to replace a real face with a genuine one. deepfake images have recently been used to malign public figures, politicians, and even average citizens. deepfake but realistic images have been used to stir political dissatisfaction, blackmail, propagate false news, and even carry out bogus terrorist attacks. Thus, identifying real images from fakes has got more challenging. To avoid these issues, this study employs transfer learning and data augmentation technique to classify deepfake images. For experimentation, 190,335 RGB-resolution deepfake and real images and image augmentation methods are used to prepare the dataset. The experiments use the deep learning models: convolutional neural network (CNN), Inception V3, visual geometry group (VGG19), and VGG16 with a transfer learning approach. Essential evaluation metrics (accuracy, precision, recall, F1-score, confusion matrix, and AUC-ROC curve score) are used to test the efficacy of the proposed approach. Results revealed that the proposed approach achieves an accuracy, recall, F1-score and AUC-ROC score of 90% and 91% precision, with our fine-tuned VGG16 model outperforming other DL models in recognizing real and deepfakes.

关键词： Deepfake detection data augmentation image processing deep learning artificial intelligence transfer learning

来源：评论

学校读者我要写书评

暂无评论

Oxide Phototransistor Array With Multiply-and-Accumulation Functions for In-Sensor image processing

引用

IEEE JOURNAL OF THE ELECTRON DEVICES SOCIETY 2025年 13卷 378-382页

作者： Wang, Saisai Jing, Xiaotao Zhang, Wanlin Wang, Rui Wang, Hong Huang, Qi Westlake Inst Optoelect Zhejiang Key Lab 3D Micro Nano Fabricat & Characte Hangzhou 311421 Peoples R China Xidian Univ Sch Microelect State Key Lab Wide Band Gap Semicond Devices & Int Xian 710071 Peoples R China

Advanced in-sensor computing paradigm has gradually become a research hotspot in this IOT era of sensor data proliferation. However, most existing in-sensor computing devices are plagued by a complex structure, and the uniformity of the sensor array is difficult to ensure. Moreover, photoconductive devices are incapable of achieving on-chip current summation because the regulation of photoresponsivity usually leads to inconsistent dark current, thereby impeding the practical implementation of artificial neural networks (ANNs) on the sensor array. In this work, we developed a 3 x 3 ultraviolet (UV) image sensor array based on solution-processed indium oxide (In2O3) phototransistors. The devices exhibit tunable responsivity and unified dark current under negative gate voltage (Vgs), enabling the direct execution of multiply-and-accumulation (MAC) operations. Consequently, two key applications of ANN were successfully demonstrated: image convolution and classification.

关键词： In-sensor computing In2O3 phototransistor tunable responsivity multiply-and-accumulation In-sensor computing In2O3 phototransistor tunable responsivity multiply-and-accumulation

来源：评论

学校读者我要写书评

暂无评论

Multifunctional In-Memory Analog-to-Digital Converter for Next-Gen Compute-in-Memory Systems

ADVANCED INTELLIGENT SYSTEMS

引用

ADVANCED INTELLIGENT SYSTEMS 2025年第5期7卷

作者： Im, Jiseong Ko, Jonghyun Hwang, Joon Kim, Jangsaeng Shin, Wonjun Koo, Ryun-Han Park, Minkyu Park, Sung-Ho Choi, Woo Young Kim, Jae-Joon Lee, Jong-Ho Seoul Natl Univ Coll Engn Interuniv Semicond Res Ctr Dept Elect & Comp Engn Seoul 08826 South Korea Sungkyunkwan Univ Dept Semicond Convergence Engn Suwon 16419 South Korea

Compute-in-memory (CIM) technology based on emerging nonvolatile memories (NVMs) has shown promise in enhancing artificial intelligence applications by integrating computation directly within NVM arrays. However, the efficiency of CIM systems is often curtailed by the substantial overhead that is caused by traditional complementary metal-oxide-semiconductor (CMOS)-based analog-to-digital converters (ADCs). Here, we report an in-memory ADC (IMADC) that leverages NVMs to perform the dual functionalities of reference generation and voltage comparison, effectively minimizing the area occupancy and energy consumption, is reported. The IMADC not only significantly outperforms traditional ADCs but also enables the inherent processing of nonlinear activation functions such as the sigmoid function, which is required for neural networks. The IMADC-based CIM system achieves software-comparable accuracy in CIFAR-10 image classification on the VGG-9 network. The IMADC exhibits significantly reduced area occupancy (45 mu m2) and energy consumption (29.6 fJ) compared to conventional CMOS-based ADCs. The IMADC, compatible with various types of NVMs, demonstrates significant potential for enhancing the efficiency of CIM systems in terms of area occupancy and energy consumption.

关键词： analog-to-digital converter compute-in-memory flash thin-film-transistor hardware-based artificial intelligence neuromorphic computing

来源：评论

学校读者我要写书评

暂无评论

SDR-Based Metal Classification Using Spectrogram images from Micro-Doppler Signatures

引用

IEEE INSTRUMENTATION & MEASUREMENT MAGAZINE 2025年第3期28卷 22-29页

作者： Liaquat, Salman Butt, Faran Awais Nasir, Faryal Aurooj Naqvi, Ijaz Haider Mahyuddin, Nor Muzlifah Muqaibel, Ali Hussein Alawsh, Saleh Ahmed Univ Sains Malaysia Elect & Elect Engn George Town Penang Malaysia Univ Management & Technol Lahore Pakistan Inst Space Technol Artificial Intelligence Islamabad Pakistan Lahore Univ Management Sci Sch Sci & Engn Lahore Pakistan Univ Sains Malaysia Sch Elect & Elect Engn George Town Malaysia King Fahd Univ Petr & Minerals Elect Engn Dept Dhahran Saudi Arabia King Fahd Univ Petr & Minerals Interdisciplinary Res Ctr Commun Syst & Sensing Dhahran Saudi Arabia

Metallic materials such as brass, copper, and aluminum are used in numerous applications, including industrial manufacturing. The vibration characteristics of these objects are unique and can be used to identify these objects from a distance. This research presents a methodology for detecting and classifying these metallic objects using the vibration dynamics induced by their micro- Doppler signatures. The proposed approach utilizes image processing techniques to extract pivotal features from spectrograms. These spectrograms originate from micro-Doppler signatures of data collected during controlled laboratory experiments where signals were transmitted towards vibrating metal sheets, and the ensuing reflections were recorded using a software-defined radio (SDR). The spectrogram data was augmented using geometric transformation to train a convolutional neural network (CNN) based machine learning model for object classification. The results indicate that the proposed CNN model achieved an accuracy of more than 95% in classifying metals into brass, copper, and aluminum. This research could be used to understand the foundations of classifying spectrogram images using micro-Doppler signatures for its applications towards enhancing the sensing capabilities in industrial and defense applications.

关键词： Vibrations Accuracy Aluminum Feature extraction Nonhomogeneous media Sensors Convolutional neural networks Copper Object recognition Spectrogram Metallic materials

来源：评论

学校读者我要写书评

暂无评论

Research on Human Resource Management and Prediction Model Based on neural Network Optimized by Genetic Algorithm 2

Research on Human Resource Management and Prediction Model B...

引用

2nd IEEE International Conference on image processing and Computer applications, ICIPCA 2024

作者： Qi, Zexuan Yang, Sihan He, Xinjian Shi, Desheng Xinjiang University of Science & Technology Xinjiang Korla China School of Information and Software Engineering University of Electronic Science and Technology of China Chengdu China School of Computer Engineering Guangzhou Huali College Guangzhou China Modern Agricultural College Honghe Vocational and Technical College Yunnan Mengzi China

ISBN: (纸本)9798350360240

This paper investigates the application of artificial neural networks optimized based on genetic algorithms in human resource management in hospital enterprises and constructs a human resource management and predictive analysis model. The model aims to support enterprises in achieving the optimal allocation of human resources by mining their internal human resource data and revealing the intrinsic correlation between personnel structure, business type and future development strategy. The study adopts multiple regression analysis method and combines the artificial neural network optimized by genetic algorithm to construct a human resource demand prediction model. Taking the operational data of Hospital X as an example, the model predicts the number of human resources in the hospital and analyzes in depth the demand for staff types in each department. The results show that the model has high accuracy and reliability in predicting the variables of hospital revenue, wage level, R&D investment and number of personnel, with a prediction error range between 0.09% and 1.27%. The research in this paper provides strong data support for enterprises, which helps them to realize the rational allocation of human resources and strategic transformation, and then maintain their competitive position in the market. © 2024 IEEE.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Compressing neural networks on Limited Computing Resources

引用

IEEE ACCESS 2025年 13卷 80063-80075页

作者： Lee, Seunghyun Lee, Dongjun Hyun, Minju Kim, Heeje Song, Byung Cheol Inha Univ Dept Elect & Comp Engn Incheon 22212 South Korea

Network compression is a crucial technique for applying deep learning models to edge or mobile devices. However, the cost of achieving higher benchmark performance through compression is continuously increasing, making network compression a significant burden-especially for small industries focused on developing compact models. Specifically, existing network compression techniques often require extensive computational resources, rendering them impractical for edge devices and small-scale applications. To democratize network compression, we propose a general-purpose framework that combines novel filter pruning and knowledge distillation techniques. First, unlike conventional filter pruning methods based on static heuristics and costly neural architecture search (NAS)-based approaches, our method leverages meta-learning for rapid and fine examination of the importance of each gate. This enables rapid and stable sub-network discovery, significantly improving the pruning process. Second, to minimize the computational cost of knowledge distillation, we introduce a synthetic teacher assistant that leverages precomputed fixed knowledge-referring to the stored feature maps/logits of the teacher network. By leveraging fixed knowledge, we mitigate the cost incurred by the teacher network and facilitate the transmission of fixed knowledge to the student via synthetic teacher assistants, thereby preventing distribution collapse. Our proposed framework dramatically reduces the compression overhead while maintaining high accuracy, achieving a 55.2% reduction in FLOPs of ResNet-50 trained on imageNet while preserving 76.2% top-1 accuracy with only 199 GPU hours-significantly lower than previous state-of-the-art methods. Overall, our framework democratizes deep learning compression by offering a cost-effective and computationally feasible solution, enabling broader adoption in low-resource environments.

关键词： Costs Knowledge engineering Logic gates Filtering algorithms Graphics processing units Computational modeling Accuracy image coding Deep learning Workstations Deep neural network compression filter pruning knowledge transfer efficient neural networks lightweight deep learning models

来源：评论

学校读者我要写书评

暂无评论

A Novel Deep Learning-Based Model for Classification of Wheat Gene Expression

引用

Computer Systems Science & Engineering 2024年第2期48卷 273-285页

作者： Amr Ismail WalidHamdy Aya MAl-Zoghby Wael AAwad Ahmed Ismail Ebada Yunyoung Nam Byeong-Gwon Kang Mohamed Abouhawwash Faculty of Science Port Said UniversityPort SaidEgypt Modern Academy for Computer Science and Management Technology Cairo11742Egypt Faculty of Computers and Artificial Intelligence Damietta UniversityNew DamiettaEgypt Department of ICT Convergence Soonchunhyang UniversityAsan31538South Korea Department of Mathematics Faculty of ScienceMansoura UniversityMansoura35516Egypt Department of Computational Mathematics Scienceand Engineering(CMSE)Michigan State UniversityEast LansingMI48824USA

Deep learning(DL)plays a critical role in processing and converting data into knowledge and *** technologies have been applied in a variety of applications,including image,video,and genome sequence *** deep learning the most widely utilized architecture is Convolutional neural networks(CNN)are taught discriminatory traits in a supervised *** comparison to other classic neural networks,CNN makes use of a limited number of artificial neurons,therefore it is ideal for the recognition and processing of wheat gene *** is an essential crop of cereals for people around the *** Genotypes identification has an impact on the possible development of many countries in the agricultural *** quantitative genetics prediction of genetic values is a central *** is an allohexaploid(AABBDD)with three distinct *** sizes of the wheat genome are quite large compared to many other kinds and the availability of a diversity of genetic knowledge and normal structure at breeding lines of wheat,Therefore,genome sequence approaches based on techniques of artificial Intelligence(AI)are *** paper focuses on using the Wheat genome sequence will assist wheat producers in making better use of their genetic resources and managing genetic variation in their breeding program,as well as propose a novel model based on deep learning for offering a fundamental overview of genomic prediction theory and current *** this paper,the hyperparameters of the network are optimized in the CNN to decrease the requirement for manual search and enhance network performance using a new proposed model built on an optimization algorithm and Convolutional neural networks(CNN).

关键词： Gene expression convolutional neural network optimization algorithm genomic prediction wheat

来源：评论

学校读者我要写书评

暂无评论

Nature and Biologically Inspired image Segmentation Techniques

引用

ARCHIVES OF COMPUTATIONAL METHODS IN ENGINEERING 2022年第3期29卷 1415-1442页

作者： Singh, Simrandeep Mittal, Nitin Thakur, Diksha Singh, Harbinder Oliva, Diego Demin, Anton Chandigarh Univ Dept Elect & Commun Engn Gharuan Punjab India Chandigarh Engn Coll Dept Elect & Commun Engn Landran Punjab India Univ Guadalajara Dept Ciencias Computac CUCEI Av Revoluc 1500 Guadalajara Jalisco Mexico Tomsk Polytech Univ Sch Comp Sci & Robot Tomsk Russia

image processing is among the significant areas of growth in the current scenario. It consist of a set of techniques typically used to enhance the raw image obtained from different scenes. Segmentation of images is an essential step in image analysis and pre-processing. During the course of the work, standard multilevel thresholding methods are very effective due to their low computational cost, reliability, reduced convergence time, and precision. Nature-inspired methods of optimization play an essential role in the processing of images. Several optimization procedures have been proposed for different image processing applications. These optimization techniques can improve the performance of image segmentation, image restoration, edge detection, image enhancement, pattern recognition, image generation, image thresholding, and image fusion algorithms. This paper includes an overview of several metaheuristic firefly algorithm (FA), differential evolution (DE), particle swarm optimization (PSO), genetic algorithm (GA), artificial bee colony optimization (ABC), etc. Moreover, artificial neural networks (ANN) and other machine learning techniques (nature or biological inspired) are discussed in context with image segmentation application and their algorithms.

关键词： image segmentation

来源：评论

学校读者我要写书评

暂无评论

Multi-Threading Method for Rapid Tool Wear Detection Based on Integrating image Classification and Object Detection

Multi-Threading Method for Rapid Tool Wear Detection Based o...

引用

2024 Conference on Spectral Technology and applications, CSTA 2024

作者： You, Zhichao Du, Yuheng Wang, Xi Li, Ziteng Liu, Huan Li, Duo Company Ltd. Shanghai200240 China Center for Precision Engineering Harbin Institute of Technology 92 West Dazhi St. Harbin150001 China Center of Ultra-Precision Optoelectronic Instrumentation Engineering Harbin Institute of Technology Harbin150001 China

ISBN: (纸本)9781510683082

This research aims to develop a multi-threading method for rapid tool wear detection by integrating image classification and object detection techniques to address the challenge of tool wear detection. The research proposes a two-stage method that leverages a fast image classification model (VGG-16) and a high-accuracy object detection model (YOLOv5) to enable efficient multi-threading detection of tool wear regions across a large number of the flank wear images. The experiment results reveal that, when the wear images account for less than 70% of the total, this method can achieve detection speeds exceeding that of YOLOv5 while maintaining comparable detection accuracy. © 2024 SPIE.

关键词： Object detection image classification image processing Data modeling Sensors artificial neural networks Convolutional neural networks Education and training Engineering Signal 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：