检索结果-内蒙古大学图书馆

ElasticDNN: On-Device neural Network Remodeling for Adapting Evolving Vision Domains at Edge

IEEE TRANSACTIONS ON COMPUTERS 2024年第6期73卷 1616-1630页

作者： Zhang, Qinglong Han, Rui Liu, Chi Harold Wang, Guoren Chen, Lydia Y. Beijing Inst Technol Beijing 100081 Peoples R China Delft Univ Technol NL-2628 Delft Netherlands

Executing deep neural networks (DNN) based vision tasks on edge devices encounters challenging scenarios of significant and continually evolving data domains (e.g. background or subpopulation shift). With limited resources, the state-of-the-art domain adaptation (DA) methods either cause high training overheads on large DNN models, or incur significant accuracy losses when adapting small/compressed models in an online fashion. The inefficient resource scheduling among multiple applications further degrades their overall model accuracy. In this paper, we present ElasticDNN, a framework that enables online DNN remodeling for applications encountering evolving domain drifts at edge. Its first key component is the master-surrogate DNN models, which can dynamically generate a small surrogate DNN by retaining and training the large master DNN's most relevant regions pertinent to the new domain. The second novelty of ElasticDNN is the filter-grained resource scheduling, which allocates GPU resources based on online accuracy estimation and DNN remodeling of co-running applications. We fully implement ElasticDNN and demonstrate its effectiveness through extensive experiments. The results show that, compared to existing online DA methods using the same model sizes, ElasticDNN improves accuracy by 23.31% and reduces adaption time by 35.67x. In the more challenging multi-application scenario, ElasticDNN improves accuracy by an average of 25.91%.

关键词： Adaptation models Training Task analysis image edge detection artificial neural networks Graphics processing units Data models Edge vision deep neural networks domain adaptation

来源：评论

学校读者我要写书评

暂无评论

3D image Analysis of Sports Technical Features and Sports Training Methods Based on artificial Intelligence

引用

JOURNAL OF TESTING AND EVALUATION 2023年第1期51卷 189-200页

作者： Mei, Zhu Shenyang Jianzhu Univ Minist Sports Shenyang 110168 Liaoning Peoples R China

This article studies the three-dimensional (3D) image analysis and sports training methods of sports technical characteristics. This research uses the current sports technology diagnostic 3D video analysis system as a platform to build a database and knowledge base based on athletes' 3D sports information and sports parameters and uses algorithms based on artificial intelligence machine-learning machines to analyze sports data, learn from it, and learn from sports technology. Actions make analytical decisions and predictions. Then, it analyzes the human-motion behavior with the concept of traditional and virtual reality technology. The effectiveness of athletes' technical movements, using mathematical statistics, artificial intelligence, and other research methods, integrates and draws on the research methods of sports biomechanics, graphical imaging, human anatomy, expert systems, and neural networks. A neural network not only inherits certain characteristics of biology but also has its own unique characteristics, such as large-scale parallel processing, strong fault tolerance, and self-learning functions. neural networks have a wide range of applications in information processing, pattern recognition, optimization, and other issues. By analyzing the application status of artificial intelligence technology in sports, the development prospects of sports training based on artificial intelligence can be inferred. Based on the acquisition of sports-related data, the evaluation of functional action modes, sports techniques, etc., is established. The multi-target feedback training method ultimately helps athletes improve their training level. Experimental data show that for the human body walking toward the camera, the rotation angle between adjacent frames is close to 0 degrees, and the translational position is basically 5 cm. The experimental results show that 3D image analysis and related sports training methods based on specific sports technical characteristics are

关键词： 3D image analysis artificial intelligence sports technical features sports training multi-target feedback training sports data

来源：评论

学校读者我要写书评

暂无评论

Integrating AI and UAV Technologies for Enhanced and Workplace Safety 27th

Integrating AI and UAV Technologies for Enhanced and Workpla...

引用

27th International conference on Interactive Collaborative Learning-ICL

作者： Badea, Cosmin Costea, Antonia Halmaghi, Alexandra Hancu, Alexandra Bogdan, Ioana Corina Modran, Horia Alexandru Transilvania Univ Brasov Dept Elect & Comp Brasov Romania

ISBN: (纸本)9783031835193;9783031835209

The integration of artificial intelligence (AI) and unmanned aerial vehicle (UAV) technologies presents a significant advancement in enhancing safety in traffic, workplace, and healthcare environments. This study explores the application of AI-driven computer vision algorithms in UAVs to detect and mitigate risks associated with substance abuse, fatigue, and health impairments. Utilizing sophisticated image processing techniques, such as edge detection and support vector machine (SVM) algorithms, drones are equipped to autonomously monitor and analyze ocular characteristics and facial expressions of individuals. The research employs a mobile phone camera and Python-based libraries to conduct real-time assessments, providing critical data to medical and industrial professionals. The study demonstrates the potential of drones to enhance safety by checking sobriety and monitoring worker health. The experimental setup includes a detailed workflow for real-time video detection and facial analysis, leveraging pre-trained models and convolutional neural networks. The results confirm the effectiveness of this approach, highlighting significant progress in AI and UAV technology. Future work aims to transition these innovations from laboratory conditions to practical, real-world applications, continuously enhancing the algorithms and expanding their applicability across various safety-critical scenarios.

关键词： Drones Surveillance Computer Vision artificial Intelligence image processing Healthcare

来源：评论

学校读者我要写书评

暂无评论

An Instrument Tone Recognition System Based on artificial Intelligence and Feature Recognition 2

An Instrument Tone Recognition System Based on Artificial In...

引用

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

作者： Hu, Zheng School of Music Zhaoqing University Guangdong Zhaoqing City526061 China

ISBN: (纸本)9798350360240

Instrument tone recognition systems have over time had the highest application value and significance in information retrieval. Notably, the traditional systems and methods often rely on convolutional neural networks (CNN) that serve as fundamental through reportedly, have a poor performance. To this effect, this work seeks to examine a resent artificial intelligence (AI) and a feature recognition system in instrument tone recognition. A proposed model in this case is integrated in an interactive system of humanoid robot. The findings of this study reveal that the proposed resent artificial intelligence (AI) and a feature recognition system in the design of an instrument tone recognition system outperforms the traditional model in terms of recognition accuracy and tonal loss rate. The findings of the research therefore providing methods and ideas that may be used in instrument tonal recognition. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Driving scene image Dehazing model based on multi-branch and multi-scale feature fusion

引用

neural networks 2025年 188卷 107495页

作者： Yin, Shi Liu, Hui Cent South Univ Sch Traff & Transportat Engn Inst Artificial Intelligence & Robot IAIR Key Lab Traff Safety TrackMinist Educ Changsha 410075 Hunan Peoples R China

image dehazing is critical for enhancing image quality in applications such as autonomous driving, surveillance, and remote sensing. This paper presents an innovative image dehazing model based on a multi-branch and multi scale feature fusion network that leverages spatial and frequency information. The model features a multi-branch architecture that combines local and global features through depthwise separable convolutions and state space models, effectively capturing both detailed and comprehensive information to improve dehazing performance. Additionally, a specialized module integrates spatial and frequency domain information by utilizing convolutional layers and Fourier transforms, enabling comprehensive haze removal through the fusion of these two domains. A feature fusion mechanism incorporates channel attention and residual connections, dynamically adjusting the importance of different channel features while preserving the global structural information of the input image. Furthermore, this is the first model to combine Mamba and convolution layers for driving scene image dehazing, achieving global feature extraction with linear complexity. Each image is processed in only 0.030 s, with a frame rate of 32.41 FPS and a processing efficiency of 67.96 MPx/s, ensuring high efficiency suitable for real-time applications. Extensive experiments on real-world foggy driving scene datasets demonstrate the superior performance of the proposed method, providing reliable visual perception capabilities and significantly improving adaptability and robustness in complex environments.

关键词： image dehazing State space model Frequency domain Autonomous driving

来源：评论

学校读者我要写书评

暂无评论

Insights Into YOLOv9-Based Breast Mass Detection Using Ultrasound images 13

Insights Into YOLOv9-Based Breast Mass Detection Using Ultra...

引用

13th International conference on image processing Theory Tools and applications

作者： Erragzi, Nouhaila Moussaid, Abdelghani Zrira, Nabila Benmiloud, Ibtissam Sebihi, Rajaa Ngote, Nabil Mohammed V Univ Fac Sci LPHE Modeling & Simulat Rabat Morocco Natl Super Sch Mines Rabat CPS2E Lab MECAtron Team Rabat Morocco Natl Super Sch Mines Rabat LISTD Lab ADOS Team Rabat Morocco Natl Super Sch Mines Rabat Abulcasis Int Univ Hlth Sci Rabat Morocco

ISBN: (纸本)9798331541859;9798331541842

The most prevalent form of cancer globally is breast cancer, which predominantly impacts women. Early detection ensures successful treatment of breast cancer, significantly improving patients' survival chances. Various imaging modalities, including mammography and ultrasound, are utilized for breast cancer screening. Incorporating new technologies is essential for better patient management, particularly for those with malignant masses. artificial intelligence can assist radiologists by training neural networks to detect breast lesions on mammograms or ultrasounds using deep learning techniques. In this article, the YOLOv9 network is trained on two public ultrasound databases, UDIAT and BUSIS. The network successfully localized malignant and benign masses with a precision of 83%, a recall of 82%, and a mAP of 87% in the UDIAT dataset. In the BUSIS dataset, our model achieved a precision of 75%, a recall of 88%, and a mAP of 90%. Furthermore, we used real Moroccan cases to evaluate the model's performance.

关键词： Ultrasound artificial Intelligence Deep Learning Breast Lesion Detection YOLOv9

来源：评论

学校读者我要写书评

暂无评论

Touching the limit of Rolling Multilayer Perceptron for efficient two-dimensional medical image segmentation

引用

ENGINEERING applications OF artificial INTELLIGENCE 2025年 153卷

作者： Liu, Yutong Zhu, Haijiang An, Ning Xu, Li Beijing Univ Chem Technol Coll Informat Sci & Technol Beijing 100029 Peoples R China China Coal Res Inst Res Inst Mine Artificial Intelligence Beijing 100013 Peoples R China State Key Lab Intelligent Coal Min & Strata Contro Beijing 100013 Peoples R China Capital Med Univ Beijing Jishuitan Hosp Dept Radiol Beijing 100035 Peoples R China

Significant progress has been made in medical image segmentation using deep learning techniques, with the Ushaped architecture being a classic choice. However, effectively capturing and integrating both local features and remote dependencies remains a key challenge for improving deep learning-based segmentation methods. In this paper, we propose a flexible Rolling Multilayer Perceptron (Rolling-MLP) module to address this issue. Building upon this concept, we present the Rolling-Unet network, which combines the strengths of Multilayer Perceptrons (MLPs) with Convolutional neural networks (CNNs) to efficiently extract and fuse local features and remote dependencies. Furthermore, to explore the potential of Rolling-MLP for two-dimensional medical image segmentation, we propose Rolling-MLP configurations with distinct receptive field shapes (linear and area-shaped) and summarize the influence of Rolling-MLP's key parameters on the shape of receptive fields. We conducted extensive experiments on four datasets, surpassing a variety of state-of-the-art methods in accuracy. Moreover, Rolling-MLP is far ahead in Central processing Unit (CPU) inference speed, indicating its potential in medical cyber-physical systems engineering applications. This paper demonstrates the strong comprehensive ability of Rolling-MLP in two-dimensional medical image segmentation tasks, providing a novel approach for constructing medical image segmentation networks, alternative to CNNs and Transformers.

关键词： Cyber-physical systems engineering application Medical image segmentation Multilayer perceptron Remote dependency

来源：评论

学校读者我要写书评

暂无评论

An EANN-Based Recommender System for Drug Recommendation 25th

An EANN-Based Recommender System for Drug Recommendation

引用

25th International conference on Engineering applications of neural networks (EANN)

作者： Al Mubasher, Hadi Awad, Mariette Amer Univ Beirut Maroun Semaan Fac Engn & Architecture Beirut Lebanon

ISBN: (纸本)9783031624940;9783031624957

Recommender systems aim to improve the user experience in a world where data and available alternatives are expanding at an unprecedented rate. Integrating Natural Language processing and artificial neural networks have resulted in better performance when compared to other recommender systems. This paper showcases the optimization of an artificial neural network-based recommender system that is used for drug recommendation, where the optimization process involves adopting ResNet-50 and a Multiple Criteria Decision Making-based recommender system to tune the learning rate of the neural network models on which the system is based. Results show that our proposed approach leads to a system that outperforms the existing similar systems.

关键词： Machine Learning Multiple Criteria Decision Making Hyperparameter Tuning Recommender Systems Natural Language processing Deep Learning artificial neural networks Drug Recommendation

来源：评论

学校读者我要写书评

暂无评论

Social Media Hate Speech Detection Using Machine Learning Algorithms: Comparative Study 5th

Social Media Hate Speech Detection Using Machine Learning Al...

引用

5th International conference on Data Science, Machine Learning and applications

作者： Dharani, P. Bagade, Nidhi Nittala, Sripriya Konkala, Sowmya Sasidhar, B. G Narayanamma Inst Technol & Sci Women Dept CSE AI&ML Hyderabad India

ISBN: (纸本)9789819780334;9789819780310;9789819780303

Cyberbullying on social media using hate speech in text is applying depreciatory dialect in message dispatches on online forums to abuse, defile, as well as ill-treat recipients. A report by the New Indian Express stated that 93% of Indian children were subordinated to cyberbullying out of which 45% were bullied by strangers and 48% were bullied by people known to them. The existing system uses feature extraction using count vectorizer with Support Vector Machine classifier to give an accuracy of 94.78%. The proposed system uses feature extraction using tokenization and padding with artificial neural networks Classifier to achieve an accuracy of 95.85%. The system examines the text content of social media dispatches using Natural Language processing through artificial neural networks.

关键词： Cyberbullying Feature extraction Count vectorizer Tokenization Padding Natural language processing artificial neural networks

来源：评论

学校读者我要写书评

暂无评论

Character Recognition Using Convolutional neural networks Based on CUDA 7

Character Recognition Using Convolutional Neural Networks Ba...

引用

7th International Youth conference on Radio Electronics, Electrical and Power Engineering, REEPE 2025

作者： Tun, Set Paing National Research University 'MPEI' Applied Mathematics and Artificial Intelligence Department Moscow Russia

ISBN: (纸本)9798331531836

The article explores character recognition using convolutional neural networks (CNNs) optimized with the CUDA platform to enhance computational efficiency. It outlines the CNN architecture, methods for leveraging GPU-based parallel data processing, and presents experimental results derived from the MNIST dataset. The study highlights that implementing CUDA drastically reduces processing time while maintaining a high level of predictive accuracy. The findings emphasize the potential of GPU acceleration in handling intensive computational tasks, making it a promising approach for real-time applications in image recognition and machine learning. © 2025 IEEE.

关键词： Convolutional neural networks

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：