检索结果-内蒙古大学图书馆

VisioSLR: A Vision Data-Driven Framework for Sign Language Video Recognition and Performance Evaluation on Fine-Tuned YOLO Models

引用

Procedia computer science 2025年 257卷 85-92页

作者： Leila Ismail Nada Shahin Henok Tesfaye Alain Hennebelle Intelligent Distributed Computing and Systems (INDUCE) Lab Department of Computer Science and Software Engineering College of Information Technology United Arab Emirates University United Arab Emirates Independent Researcher

With the emergence of AI for good, there has been an increasing interest in building computer vision data-driven deep learning inclusive AI solutions. Sign language Recognition (SLR) has gained attention recently. It is an essential component of a sign-to-text translation system to support the deaf and hard-of-hearing population. This paper presents a computer VISIOn data-driven deep learning framework for Sign Language video Recognition (VisoSLR). VisioSLR provides a precise measurement of translating signs for developing an end-to-end computational translation system. Considering the scarcity of sign language datasets, which hinders the development of an accurate recognition model, we evaluate the performance of our framework by fine-tuning the very well-known YOLO models, which are built from a signs-unrelated collection of images and videos, using a small-sized sign language dataset. Gathering a sign language dataset for signs training would involve an enormous amount of time to collect and annotate videos in different environmental setups and multiple signers, in addition to the training time of a model. Numerical evaluations of VisioSLR show that our framework recognizes signs with a mean average precision of 97.4%, 97.1%, and 95.5% and 11, 12, and 12 milliseconds of recognition time on YOLOv8m, YOLOv9m, and YOLOv11m, respectively.

关键词： Artificial intelligence computer vision Deep learning Fine-tuning Machine translation Neural machine translation Sign language Sign language recognition Transfer learning YOLO

来源：评论

学校读者我要写书评

暂无评论

YOLO-RAW: Advancing UAV Detection With Robustness to Adverse Weather Conditions

引用

IEEE Transactions on intelligent Transportation systems 2025年第6期26卷 7857-7873页

作者： Munir, Adnan Siddiqui, Abdul Jabbar Hossain, M. Shamim El-Maleh, Aiman Computer Engineering Department Dhahran31261 Saudi Arabia KFUPM SDAIA-KFUPM Joint Research Center for Artificial Intelligence IRC for Intelligent Secure Systems and Computer Engineering Department Dhahran31261 Saudi Arabia King Saud University College of Computer and Information Sciences Research Chair of Pervasive and Mobile Computing Department of Software Engineering Riyadh12372 Saudi Arabia KFUPM Computer Engineering Department Information and Computer Science Department IRC for Intelligent Secure Systems Dhahran31261 Saudi Arabia

With the widespread adoption of unmanned aerial vehicles (UAVs) in various applications (e.g., aerial transportation, traffic monitoring), there have been apprehensions regarding the associated risks of employing UAVs in both civilian and military contexts, including concerns about privacy infringement, safety issues, and security threats. Although several methods have been proposed to detect UAVs, a pressing open challenge is posed by varying adverse weather conditions that could degrade the performance of many existing methods. To address these limitations, this work proposes a YOLO-based (You Only Look Once) novel model, YOLO-RAW that exhibits improved performance in adverse weather conditions and considers different scales of UAVs. Furthermore, to facilitate a more comprehensive evaluation of the proposed model’s effectiveness in UAV detection, we have curated a complex background dataset and introduced three distinct test sets affected by adverse weather conditions. These three test sets comprise the Rainy Test Set (RTS), the AWGN (Additive White Gaussian Noise) Test Set (ATS), and the Motion Blurred Test Set (MBTS). The comprehensive experiments demonstrate the effectiveness of the proposed YOLO-RAW model over its counterparts in detecting UAVs under adverse conditions. © 2000-2011 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Guest Editorial for Selected Papers From BIOKDD 2023

IEEE Transactions on Computational Biology and Bioinformatic...

引用

IEEE Transactions on Computational Biology and Bioinformatics 2025年第2期22卷 394-396页

作者： Da Yan Ariful Azad Jie Hou Jake Y. Chen Department of Computer Science Indiana University Bloomington Bloomington IN USA Department of Intelligent Systems Engineering Indiana University Bloomington Bloomington IN USA Department of Computer Science Saint Louis University St. Louis MO USA Department of Biomedical Informatics and Data Science University of Alabama at Birmingham Birmingham AL USA

来源：评论

学校读者我要写书评

暂无评论

Model for Anomaly Detection in Reducing Water Losses in the Distribution Network of Peru using Machine Learning Algorithms

Model for Anomaly Detection in Reducing Water Losses in the ...

引用

Software Engineering and computer science (CSECS), International Conference on

作者： Mayra Rivera Pricila Lucero José Luis Castillo-Sequera Lenis Wong Information Systems Engineering Program Universidad Peruana de Ciencias Aplicadas Lima Peru Department of Computer Science Group: Health Computing and Intelligent System Universidad de Alcalá Alcalá de Henares Spain

ISBN: (数字)9798331522216

ISBN: (纸本)9798331522223

Water leakage in distribution networks is a significant challenge, especially in regions with limited infrastructure like Huancayo, Peru, where losses account for 32.82% of the distributed volume. This study introduces a machine learning-based approach to detect leaks using four algorithms: Autoencoder LSTM, Isolation Forest, One-Class SVM, and K-Nearest Neighbors (KNN). The methodology involved preprocessing historical consumption data (2018–2024) into 12-month temporal sequences per client and evaluating the models based on F1 Score, Precision, and Mean Absolute Error (MAE). Among the algorithms, the Autoencoder LSTM demonstrated superior performance with the highest precision (0.89) and the lowest MAE (0.00402). Its robustness in high-variability contexts enables early and reliable leak detection, providing a cost-effective solution for optimizing water management in resource-constrained environments.

关键词： Support vector machines Machine learning algorithms Autoencoders Distribution networks Forestry Nearest neighbor methods Robustness Leak detection Long short term memory Water resources

来源：评论

学校读者我要写书评

暂无评论

Automated Parts Segmentation of Sperm via a Contrastive Learning-Based Part Matching Network

引用

IEEE Transactions on Automation science and Engineering 2025年 22卷 15314-15325页

作者： Chen, Wenyuan Song, Haocong Shan, Guanqiao Dai, Changsheng Liu, Hang Jiang, Aojun Sun, Chen Ru, Changhai Librach, Clifford Zhang, Zhuoran Sun, Yu University of Toronto Department of Computer Science TorontoONM5S 2E4 Canada University of Toronto Department of Mechanical and Industrial Engineering TorontoONM5S 3G8 Canada Dalian University of Technology Institute of Robotics and Intelligent Systems Liaoning Dalian China Suzhou University of Science and Technology School of Electronic and Information Engineering Suzhou215009 China CReATe Fertility Centre TorontoONM5G 1N8 Canada

Sperm morphology measurement is vital for diagnosing male infertility, which involves quantification of multiple subcellular parts for each sperm. Instance-aware part segmentation networks have been introduced to address this task by automatically identifying individual sperm and segmenting their subcellular parts. However, major limitations of state-of-the-art instance-aware part segmentation networks include: 1) they are time-consuming and computational expensive due to sequential processing and multi-stage frameworks;2) they perform poorly for densely packed sperm that overlap or cross over one another. To overcome these challenges, this paper proposes 1) integrating instance identification and subcellular part segmentation within a single-stage framework to save inference time and memory usage;2) dividing a sperm target into simpler components (head and tail) to improve prediction accuracy, followed by a contrastive learning-based matching method to pair the head and tail. Experimental results on our clinically collected human sperm dataset demonstrated that the proposed network not only outperformed state-of-the-art CP-Net (by 3.5% APpvol) but also achieved realtime inference (48.0 frames per second), effectively meeting the clinical requirements for automated parts segmentation of sperm. © 2004-2012 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Dynamic Intent Queries for Motion Transformer-based Trajectory Prediction

arXiv

引用

arXiv 2025年

作者： Demmler, Tobias Hartung, Lennart Tamke, Andreas Dang, Thao Hegai, Alexander Haug, Karsten Mikelsons, Lars Robert Bosch GmbH Stuttgart Germany Institute for Intelligent Systems Department of Computer Science and Engineering Esslingen University Germany University of Augsburg Germany

In autonomous driving, accurately predicting the movements of other traffic participants is crucial, as it significantly influences a vehicle’s planning processes. Modern trajectory prediction models strive to interpret complex patterns and dependencies from agent and map data. The Motion Transformer (MTR) architecture and subsequent work define the most accurate methods in common benchmarks such as the Waymo Open Motion Benchmark. The MTR model employs pre-generated static intention points as initial goal points for trajectory prediction. However, the static nature of these points frequently leads to misalignment with map data in specific traffic scenarios, resulting in unfeasible or unrealistic goal points. Our research addresses this limitation by integrating scene-specific dynamic intention points into the MTR model. This adaptation of the MTR model was trained and evaluated on the Waymo Open Motion Dataset. Our findings demonstrate that incorporating dynamic intention points has a significant positive impact on trajectory prediction accuracy, especially for predictions over long time horizons. Furthermore, we analyze the impact on ground truth trajectories which are not compliant with the map data or are illegal maneuvers. Copyright © 2025, The Authors. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Cardiac MRI Segmentation of Ventricular Structures and Myocardium Using U-Net Variants

Cardiac MRI Segmentation of Ventricular Structures and Myoca...

引用

Advanced Research in Computing (ICARC), International Conference on

作者： Chinthalanka B. Wijesinghe Dulani Meedeniya Pratheepan Yogarajah Department of Computer Science and Engineering University of Moratuwa Sri Lanka School of Computing Engineering and Intelligent Systems Ulster University Londonderry United Kingdom

ISBN: (数字)9798331530983

ISBN: (纸本)9798331530990

Accurate segmentation of the ventricular structures and myocardium from Cardiac Magnetic Resonance (CMR) images is essential to diagnose and manage cardiovascular diseases. This study systematically evaluates the performance of five U-Net variants in cardiac MRI segmentation using the Automated Cardiac Diagnosis Challenge (ACDC) dataset and a hybrid loss function combining Cross-Entropy and dice losses. Among the variants, the Feature Pyramid U-Net achieved the best performance, with Dice coefficients of 0.9388 (Left Ventricle), 0.8759 (Right Ventricle), and 0.8426 (Myocardium), showcasing its superior ability to capture multi-scale features and segment complex anatomical structures. The comprehensive and standardized evaluation conducted in this study provides valuable insights into the strengths and limitations of these architectures for cardiac segmentation.

关键词： Deep learning Image segmentation Magnetic resonance imaging Computational modeling Magnetic resonance computer architecture Myocardium Cardiovascular diseases Artificial intelligence Next generation networking

来源：评论

学校读者我要写书评

暂无评论

PSA: DIFFERENTIALLY PRIVATE STEERING FOR LARGE LANGUAGE MODEL ALIGNMENT

arXiv

引用

arXiv 2025年

作者： Goel, Anmol Hu, Yaxi Gurevych, Iryna Sanyal, Amartya Technical University of Darmstadt Germany Max Planck Institute for Intelligent Systems Tübingen Germany Department of Computer Science University of Copenhagen Denmark

Aligning Large Language Models (LLMs) with human values and away from undesirable behaviors (such as hallucination) has become increasingly important. Recently, steering LLMs towards a desired behavior via activation editing has emerged as an effective method to mitigate harmful generations at inference-time. Activation editing modifies LLM representations by preserving information from positive demonstrations (e.g., truthful) and minimising information from negative demonstrations (e.g., hallucinations). When these demonstrations come from a private dataset, the aligned LLM may leak private information contained in those private samples. In this work, we present the first study of aligning LLM behavior with private datasets. Our work proposes the Private Steering for LLM Alignment (PSA) algorithm to edit LLM activations with differential privacy (DP) guarantees. We conduct extensive experiments on seven different benchmarks with open-source LLMs of different sizes (0.5B to 7B) and model families (LlaMa, Qwen, Mistral and Gemma). Our results show that PSA achieves DP guarantees for LLM alignment with minimal loss in performance, including alignment metrics, open-ended text generation quality, and general-purpose reasoning. We also develop the first Membership Inference Attack (MIA) for evaluating and auditing the empirical privacy for the problem of LLM steering via activation editing. Our experiments support the theoretical guarantees by showing improved guarantees for our PSA algorithm compared to several existing non-private techniques.1 © 2025, CC BY.

关键词： Demonstrations

来源：评论

学校读者我要写书评

暂无评论

Risk-Aware Sensitive Property-Driven Resource Management in Cloud Datacenters

arXiv

引用

arXiv 2025年

作者： Felemban, Muhamad Almutairi, Abdulrahman Ghafoor, Arif Computer Engineering Department The Information and Computer Science Department The Interdisciplinary Research Center for Intelligent Secure Systems KFUPM Saudi Arabia Purdue University West LafayetteIN United States Computer Engineering Department The College of Computer and Information Science in King Saud University Riyadh Saudi Arabia

Organizations are increasingly moving towards the cloud computing paradigm, in which an on-demand access to a pool of shared configurable resources is provided. However, security challenges, which are particularly exacerbated by the multitenancy and virtualization features of cloud computing, present a major obstacle. In particular, sharing of resources among potentially untrusted tenants in access controlled cloud datacenters can result in increased risk of data leakage. To address such risk, we propose an efficient risk-aware sensitive property-driven virtual resource assignment mechanism for cloud datacenters. We have used two information-theoretic measures, i.e., KL-divergence and mutual information, to represent sensitive properties in the dataset. Based on the vulnerabilities of cloud architecture and the sensitive property profile, we have formulated the problem as a cost-drive optimization problem. The problem is shown to be NP-complete. Accordingly, we have proposed two heuristics and presented simulation based performance results for cloud datacenters with multiple sensitivity. © 2025, CC BY.

关键词： Cloud platforms

来源：评论

学校读者我要写书评

暂无评论

Evaluating Multi-Modal LLMs for Automatically Recognizing Semantic Elements in UML Use Case Diagram Images

Evaluating Multi-Modal LLMs for Automatically Recognizing Se...

引用

IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER)

作者： Jameleddine Hassine Information and Computer Science Department KFUPM Dhahran Saudi Arabia Interdisciplinary Research Center for Intelligent Secure Systems KFUPM Dhahran Saudi Arabia

ISBN: (数字)9798331535100

ISBN: (纸本)9798331535117

Requirements engineering commonly employs UML Use Case Diagrams (UCD) to visually capture system interactions and functionality, facilitating clear communication between stakeholders. Recognizing and extracting semantic information from UCDs is essential for applications such as automated requirements extraction and system design validation, which improves software analysis accuracy, and streamlines model understanding for both developers and stakeholders. Recent advancements in large language models (LLMs) with visual processing capabilities enable interpreting intricate diagrammatic content. This paper evaluates multi-modal LLMs, specifically GPT-4o and GPT-4o-mini, in accurately identifying semantic elements within UCDs. We conducted experiments on a new dataset of UCDs and other diagrams collected from online sources. Experimental results show that both models struggled to accurately identify and interpret key UCD elements, often misclassifying or overlooking essential ones.

关键词： Visualization Image recognition Large language models Unified modeling language Semantics Software Stakeholders Requirements engineering Data mining System analysis and design

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：