检索结果-内蒙古大学图书馆

Spatiotemporal crowds features extraction of infrared images using neural network

Journal of Ambient Intelligence and Humanized Computing 2024年第4期15卷 2543-2556页

作者： Al-Oraiqat, Anas M. Drieiev, Oleksandr Drieieva, Hanna Meleshko, Yelyzaveta AlRawashdeh, Hazim Al-Oraiqat, Karim A. Hasan, Yassin M. Y. Maricar, Noor Khan, Sheroz Department of Computer Science College of Engineering and Information Technology Onaizah Colleges Qassim Saudi Arabia Department of Cybersecurity and Software Central Ukrainian National Technical University Kropyvnytskyi Ukraine Department of Artificial Intelligence Faculty of Information Technology Zarqa University Zarqa Jordan Department of Electrical Engineering College of Engineering and Information Technology Onaizah Colleges Qassim Saudi Arabia

Crowds can lead up to severe disasterous consequences resulting in fatalities. Videos obtained through public cameras or captured by drones flying overhead can be processed with artificial intelligence-based crowd analysis systems. Being a hot area of research over the past few years, the goal is not only to identify the presence of crowds but also to predict the probability of crowd-formation in order to issue timely warnings and preventive measures. Such systems will significantly reduce the probablity of the potential disasters. Developing effective systems is a challenging task, especially due to factors such as naturally occuring diverse conditions, variations in people or background pixel areas, noise, behaviors of individuals, relative amounts/distributions/directions of crowd movements, and crowd building reasons. This paper proposes an infrared video processing system based on U-Net convolutional neural network for crowd monitoring in infrared video frames to help estimate the people crowd with normal or abnormal trends. The proposed U-Net architecture aims to efficiently extract crowd features, achieve sufficient people marking-up accuracy, competitively with optimal network configurations in terms of the depth and number of filters to consequently minimise the number of coefficients. For further faster processing, hardware resources/implementation area savings, and lower power, the optimized network coefficients measured are represented in Canonic-Signed Digit with minimal number of nonzero (± 1) digits, minimizing the number of underlying shift-add/subtract operations of all multipliers. The achieved significantly reduced computational cost makes the proposed U-Net effectively suitable for resource-constrained and low power applications. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

OLLIE: Imitation Learning from Offline Pretraining to Online Finetuning 41

OLLIE: Imitation Learning from Offline Pretraining to Online...

引用

41st International Conference on Machine Learning, ICML 2024

作者： Yue, Sheng Hua, Xingyuan Ren, Ju Lin, Sen Zhang, Junshan Zhang, Yaoxue Department of Computer Science and Technology Tsinghua University Beijing China Zhongguancun Laboratory Beijing China Department of Computer Science University of Houston Texas United States Department of Electrical and Computer Engineering University of California Davis United States

In this paper, we study offline-to-online Imitation Learning (IL) that pretrains an imitation policy from static demonstration data, followed by fast finetuning with minimal environmental interaction. We find the naïve combination of existing offline IL and online IL methods tends to behave poorly in this context, because the initial discriminator (often used in online IL) operates randomly and discordantly against the policy initialization, leading to misguided policy optimization and unlearning of pretraining knowledge. To overcome this challenge, we propose a principled offline-to-online IL method, named OLLIE, that simultaneously learns a near-expert policy initialization along with an aligned discriminator initialization, which can be seamlessly integrated into online IL, achieving smooth and fast finetuning. Empirically, OLLIE consistently and significantly outperforms the baseline methods in 20 challenging tasks, from continuous control to vision-based domains, in terms of performance, demonstration efficiency, and convergence speed. This work may serve as a foundation for further exploration of pretraining and finetuning in the context of IL. Copyright 2024 by the author(s)

关键词： Contrastive Learning

来源：评论

学校读者我要写书评

暂无评论

Enhancing Weakly Supervised Medical Segmentation via Heterogeneous Co-training with Box-Wise Augmentation and Pseudo-Label Filtering 6th

Enhancing Weakly Supervised Medical Segmentation via Hetero...

引用

6th IFIP TC 12 International Conference on Intelligence science, ICIS 2024

作者： Wang, You Qi, Lei Yu, Qian Shi, Yinghuan Gao, Yang State Key Laboratory of Novel Software Technology National Insititute of Health-care Data Science Nanjing University Nanjing China School of Computer Science and Engineering Southeast University Nanjing China School of Data and Computer Science Shandong Woman’s University Jinan China

ISBN: (纸本)9783031712524

In this paper, we introduce an innovative approach to weakly supervised medical image segmentation with box annotations. Different from the previous methods which simply utilize a single conventional network with the same augmentation techniques widely used in supervised segmentation, we aim to introduce diverse augmentations and heterogenous networks to leverage the box annotations for promising generalization ability. Specifically, to amplify the diversity between the contents within the box and its surroundings, we propose the interior and exterior box augmentation (IEBA) technique, in which distinct augmentation techniques are employed for regions inside and outside the bounding boxes. Also, for the purpose of selecting pseudo-labels of superior quality, we propose the pseudo-label filter module (PLFM) to eliminate unreliable pseudo-labels. Besides, as CNN demonstrates superior capabilities in acquiring local information, and ViT specializes in capturing global context, we facilitate a bidirectional learning process between CNN and ViT through quadruple cross consistency losses (QCCL). In inference, we only employ the superior model from the validation set to obtain parameter efficiency. Our approach is evaluated across four tasks on two public datasets, utilizing the 3D dice similarity coefficient as the evaluation metric. The experimental results show that the proposed method outperforms the state-of-the-art comparison methods. © IFIP International Federation for Information Processing 2025.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Predicting the Unseen: A Novel Dataset for Hidden Intention Localization in Pre-abnormal Analysis 24

Predicting the Unseen: A Novel Dataset for Hidden Intention ...

引用

32nd ACM International Conference on Multimedia, MM 2024

作者： Qi, Zehao Zhang, Ruixu Hu, Xinyi Liu, Wenxuan Wang, Zheng National Engineering Research Center for Multimedia Software Institute of Artificial Intelligence Wuhan University Hubei Key Laboratory of Multimedia and Network Communication Engineering School of Computer Science Wuhan China School of Computer Science and Artificial Intelligence Wuhan University of Technology Wuhan China

ISBN: (纸本)9798400706868

Our paper introduces a novel video dataset specifically for Temporal Intention Localization (TIL), aimed at identifying hidden abnormal intention in densely populated and complex environments. Traditional Temporal Action Localization (TAL) frameworks, focusing on overt actions within constrained temporal intervals, often miss subtle pre-abnormal actions that unfold over extended periods. Our dataset comprises 228 videos with 5790 clips, each annotated to capture fine-grained actions within ambiguous temporal boundaries using the Joint-Linear-Assignment methodology. This approach enables detailed analysis of the evolution of abnormal intention over time. To detect subtle, hidden intention, we developed the Intention-Action Fusion module, an creative approach integrating dynamic feature fusion across 11 behavioral subcategories, significantly enhancing the model's ability to discern nuanced intention. This enhancement has led to performance improvements of up to 139% in specific scenarios, dramatically boosting the model's sensitivity and interpretability, crucial for advancing proactive surveillance systems. By pushing the boundaries of technology, our dataset and methodologies foster proactive surveillance systems capable of preemptively identifying potential threats from nuanced behavioral patterns, encouraging further exploration into the complexities of intention beyond observable actions. The dataset is available at https://***/Zzz99999/Hidden-Abnormal-Intention. © 2024 ACM.

关键词： Security systems

来源：评论

学校读者我要写书评

暂无评论

Secure speaker identification in open and closed environments modeled with symmetric comb filters

引用

Multimedia Tools and Applications 2025年第18期84卷 19147-19189页

作者： Shafik, Amira Monir, Mohamad El-Shafai, Walid Khalaf, Ashraf A. M. Nassar, M.M. El-Fishawy, Adel S. El-Din, M. A. Zein Dessouky, Moawad I. El-Rabaie, El-Sayed M. Abd El-Samie, Fathi E. Department Electronics and Electrical Communications Engineering Faculty of Electronic Engineering Menoufia University Menouf32952 Egypt Security Engineering Laboratory Department of Computer Science Prince Sultan University Riyadh11586 Saudi Arabia Electrical Engineering Department Faculty of Engineering Minia University Minia61519 Egypt Department of Information Technology College of Computer and Information Sciences Princess Nourah bint Abdulrahman University P.O. Box 84428 Riyadh11671 Saudi Arabia

Speech is a fundamental means of human interaction. Speaker Identification (SI) plays a crucial role in various applications, such as authentication systems, forensic investigation, and personal voice assistance. However, achieving robust and secure SI in both open and closed environments remains challenging. To address this issue, researchers have explored new techniques that enable computers to better understand and interact with humans. Smart systems leverage Artificial Neural Networks (ANNs) to mimic the human brain in identifying speakers. However, speech signals often suffer from interference, leading to signal degradation. The performance of a Speaker Identification System (SIS) is influenced by various environmental factors, such as noise and reverberation in open and closed environments, respectively. This research paper is concerned with the investigation of SI using Mel-Frequency Cepstral Coefficients (MFCCs) and polynomial coefficients, with an ANN serving as the classifier. To tackle the challenges posed by environmental interference, we propose a novel approach that depends on symmetric comb filters for modeling. In closed environments, we study the effect of reverberation on speech signals, as it occurs due to multiple reflections. To address this issue, we model the reverberation effect with comb filters. We explore different domains, including time, Discrete Wavelet Transform (DWT), Discrete Cosine Transform (DCT), and Discrete Sine Transform (DST) domains for feature extraction to determine the best combination for SI in case of reverberation environments. Simulation results reveal that DWT outperforms other transforms, leading to a recognition rate of 93.75% at a Signal-to-Noise Ratio (SNR) of 15 dB. Additionally, we investigate the concept of cancelable SI to ensure user privacy, while maintaining high recognition rates. Our simulation results show a recognition rate of 97.5% at 0 dB using features extracted from speech signals and their DCTs. Fo

关键词： Speech enhancement

来源：评论

学校读者我要写书评

暂无评论

Artificial Intelligence For Realtime Face Recognition Attendance Using College Classrooms and Buses

Artificial Intelligence For Realtime Face Recognition Attend...

引用

2024 International Conference on Electrical, Electronics and Computing Technologies, ICEECT 2024

作者： Ahila, A. Hosanna Princye, P. Poonguzhali, A. Kavivendhan Deepa, M. Arthy, A. Saravanakumar, R. Sri Sairam College of Engineering Department of Electronics and Communication Engineering Anekal Bangalore India Francis Xavier Engineering College Department of Computer Science and Business Systems Tamil Nadu Tirunelveli India Iconix Software Solution Software Analyst Tamil Nadu Tirunelveli India

ISBN: (纸本)9798350378092

Teachers take attendance by having pupils sign in or check-in classes and transportation. Student absences often result from individual mistakes. This article examines a technology that records data from classroom photographs of every student's face. This research uses an Adaptive Boost Classifier, Random Forest (RF), and Deep Convolutional Neural Networks (DCNNs). The model performs well on the DCNN model with 88 and 92% accuracy and on the ResNet50 pre-trained model with 97.21% accuracy. After detecting each student's face, they recorded their present status in an Excel document. It kept the best system implementation approach based on performance. © 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

TTAG+R: A Dataset of Google Play Store's Top Trending Android Games and User Reviews 22

TTAG+R: A Dataset of Google Play Store's Top Trending Androi...

引用

22nd IEEE International Conference on software Quality, Reliability and Security Companion, QRS-C 2022

作者： Chand, Raheela Rehman Khan, Saif Ur Hussain, Shahid Wang, Wenli Department of Computer Science Islamabad Pakistan School of Engineering Penn State University Department of Computer Science and Software Engineering United States

ISBN: (纸本)9798350319910

Context: Android games are gaining wide attention from users in recent years. However, the existing literature reports alarming statistics about banning popular and top-trending Android apps. The popular gaming apps have been removed from Google Play Store due to various user concerns. Objectives: The goal of this work is twofold: (i) to assist the researchers and practitioners in identifying the state-of-the-art challenges, constraints, and compliments about Android apps for future Android-specific studies, and (ii) to encourage active users' perspectives on the Android development process because usability remains a core deciding factor about the success or failure of Android apps. Method: To accomplish this, we introduce a novel open-source dataset, Top Trending Android apps with their user Reviews (TTAG+R) in GitHub. Results and Contributions: Briefly, TTAG+R presents information about 245 top trending Android Free Games, 97 top trending Android Grossing Games, and 52 top trending Android Paid Games with a total of 8,423 user reviews in 12 different. csv files. The main contributions of this paper are: (i) provides one-place comprehensive data on Android Apps, (ii) describes various features of Android apps and their user reviews, (iii) reports the updated and latest knowledge about Android apps, and (iv) provides the data in an unfiltered form so that researchers may not find difficulty in using this dataset in their data-driven experimentation. From a research implication viewpoint, the dataset supports: (i) understanding the usability characteristics of Android apps, (ii) discovering current trends and pitfalls in Android apps, and (iii) analyzing the Android financial market. Conclusion and Future Work: Thus, TTAG+R is freely available to the research community, and useful for future enhancements in the Android domain. In the future, we plan to keep the data up-to-date with the most recent information for the continued usage of the dataset. © 2022 IEEE.

关键词： Android (operating system)

来源：评论

学校读者我要写书评

暂无评论

A Review of Blockchain in Internet of Medical Things

A Review of Blockchain in Internet of Medical Things

引用

International Conference on. Cryptology and Network Security with Machine Learning, ICCNSML 2023

作者： Mansouri, Houssem Hireche, Rachida Benrebbouh, Chahrazed Pathan, Al-Sakib Khan LRSD Laboratory Department of Computer Science Faculty of Sciences Setif 1 University - Ferhat Abbas Sétif Algeria Department of Computer Science and Engineering United International University Dhaka Bangladesh

ISBN: (纸本)9789819706402

During the past few years, especially after the emergence of the Covid-19 pandemic, researchers have devoted their efforts in improving the global health sector by supporting it with the latest technologies. Among these technologies, we often hear about Internet of Medical Things (IoMT) and blockchain in the effort of facilitating the patients and medical staff to preserve their security and confidentiality of the patient data and protect it from every hacking attempt to steal or falsify. In fact, during just the last three years alone, dozens of new schemes have been proposed in the literature for the integration of blockchain with IoMT for the healthcare field. Therefore, we present in this paper a review of some the notable works in this area. Our intent is to explain the basic principles of this field and classify the notable proposed schemes, as this study suggests potentially interesting avenues for future research to use it as a reference material by the researchers in this field. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Health care

来源：评论

学校读者我要写书评

暂无评论

Federated Non-Intrusive Load Monitoring for Smart Homes Utilizing Attention-Based Aggregation

Federated Non-Intrusive Load Monitoring for Smart Homes Util...

引用

2023 IEEE International Conference on Industry 4.0, Artificial Intelligence, and Communications Technology, IAICT 2023

作者： Kaspour, Shamisa Yassine, Abdulsalam Lakehead University Department of Computer Science Thunder Bay Canada Lakehead University Department of Software Engineering Thunder Bay Canada

ISBN: (纸本)9798350313635

Nowadays, Non-Intrusive Load Monitoring (NILM) with Federated Learning (FL) framework has become a growing study towards providing a secure energy disaggregation system in smart homes. This study aims at deploying an attention-based aggregation (FedAtt) approach in FL to emphasize agents' behavioral differences when consuming energy from various appliances. The goal of the proposed technique is to minimize the weighted distance between the parameters of the local model and the global model to better represent each local model's characteristics. In this paper, we examine two different models for NILM: Short Sequence-to-Point (SS2P) and Variational Auto-Encoder (VAE). Our goal is to evaluate the effectiveness of FedAtt. The evaluation of the framework was carried out using the UK-DALE and REFIT datasets. The obtained results were then compared against centralized approaches of the models as well as FedAvg. Our findings show that FedAtt generates comparable results to the centralized model and FedAvg while improving the stability of FL at different values of added noise to local parameters. © 2023 IEEE.

关键词： Automation

来源：评论

学校读者我要写书评

暂无评论

Phase Re-Service in Reinforcement Learning Traffic Signal Control 27

Phase Re-Service in Reinforcement Learning Traffic Signal Co...

引用

27th IEEE International Conference on Intelligent Transportation Systems, ITSC 2024

作者： Zhang, Zhiyao Gunter, George Quinones-Grueiro, Marcos Zhang, Yuhang Barbour, William Biswas, Gautam Work, Daniel Institute for Software Integrated Systems Vanderbilt University Department of Civil and Environmental Engineering United States Institute for Software Integrated Systems Vanderbilt University United States Institute for Software Integrated Systems Vanderbilt University Department of Computer Science United States

ISBN: (纸本)9798331505929

This article proposes a novel approach to traffic signal control that combines phase re-service with reinforcement learning (RL). The RL agent directly determines the duration of the next phase in a pre-defined sequence. Before the RL agent's decision is executed, we use the shock wave theory to estimate queue expansion at the designated movement allowed for re-service and decide if phase re-service is necessary. If necessary, a temporary phase re-service is inserted before the next regular phase. We formulate the RL problem as a semi-Markov decision process (SMDP) and solve it with proximal policy optimization (PPO). We conducted a series of experiments that showed significant improvements thanks to the introduction of phase re-service. Vehicle delays are reduced by up to 29.95% of the average and up to 59.21% of the standard deviation. The number of stops is reduced by 26.05% on average with 45.77% less standard deviation. © 2024 IEEE.

关键词： Markov processes

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：