检索结果-内蒙古大学图书馆

Texture and Orientation-based Feature Extraction for Robust Facial Expression Recognition

学校读者我要写书评

暂无评论

Texture and Orientation-based Feature Extraction for Robust ...

International Conference on Software engineering Research, Management and Applications (SERA)

作者： Sajid Ali Khan Mohammed Bin Abdulrahman Alawairdhi Mousa AL-Akhras Department of Software Engineering Foundation University Islamabad Islamabad Pakistan College of Computing and Informatics Saudi Electronic University Saudi Arabia Computer Information Systems Department King Abdullah II School for Information Technology The University of Jordan Amman

Facial expressions are the most effective way to characterize people’s motives, emotions, and feelings. Several new methods are proposed each year; however, the accuracy of facial expression recognition still needs to be improved especially in uncontrolled conditions. In this paper, we propose a hybrid facial expression model that considers both texture and orientation features to classify expressions. Two types of descriptors namely Local binary pattern and Weber local descriptor are used to preserve the local intensity information and orientation of edges. In the next step, computing the Histograms of oriented gradients (HOG) features from the Local binary pattern and Weber local descriptor images to capture micro-expressions. Then, the AdaBoost feature selection algorithm is utilized to choose the best features from the combined HOG features. The results of the experiments demonstrate that the method proposed in this study performs better than existing methods.

关键词：

Smart Traffic Light Using Raspberry Pi and Digital Image Processing

学校读者我要写书评

暂无评论

Smart Traffic Light Using Raspberry Pi and Digital Image Pro...

International Conference on Wireless and Telematics (ICWT)

作者： Helfy Susilawati Shelmy Luthfi Pathur Rahman Ade Rukmana Iik Muhamad Malik Matin Sarbini Nanang Ismail Electrical Engineering Departement Universitas Garut Garut Indonesia Computer and Informatics Engineering Jakarta State Polytechnic Jakarta Indonesia Psychology Department UIN Sunan Gunung DJati Bandung Bandung Indonesia Departement of Electrical Engineering UIN Sunan Gunung DJati Bandung Bandung Indonesia

Congestion is one of the problems in traffic, especially at intersections with traffic lights. In this study, a traffic light management system was developed by utilizing image processing that can adjust the duration in the queue depending on road conditions. The system is developed using Raspberry Pi 4 as the controller. The system will capture the image of the traffic conditions via cameras and then perform the image processing using OpenCV and Tensorflow to calculate the number of vehicles. The number of vehicles will be sent to the webserver database to be processed as the input for traffic system settings. The testing shows that the system can set the timing of the traffic lights. When there are 3 vehicles in the queue with traffic light is red then the system set the timing for the green light to 10 seconds. Additional of 1 vehicle in the queue will increase the duration of the green light by 4 seconds.

关键词：

$\mathop {\mathtt {HAM}}$HAM: Hybrid Associations Models for Sequential Recommendation

学校读者我要写书评

暂无评论

IEEE Transactions on Knowledge and Data engineering 2021年第10期34卷 4838-4853页

作者： Bo Peng Zhiyun Ren Srinivasan Parthasarathy Xia Ning Department of Computer Science and Engineering The Ohio State University Columbus OH USA Department of Biomedical Informatics The Ohio State University Columbus OH USA Department of Biomedical Informatics Department of Computer Science and Engineering Translational Data Analytics Institute The Ohio State University Columbus OH USA

Sequential recommendation aims to identify and recommend the next few items for a user that the user is most likely to purchase/review, given the user's purchase/rating trajectories. It becomes an effective tool to help users select favorite items from a variety of options. In this manuscript, we developed hybrid associations models ( $\mathop {\mathtt {HAM}}\limits$ ) to generate sequential recommendations using three factors: 1) users’ long-term preferences, 2) sequential, high-order and low-order association patterns in the users’ most recent purchases/ratings, and 3) synergies among those items. $\mathop {\mathtt {HAM}}\limits$ uses simplistic pooling to represent a set of items in the associations, and element-wise product to represent item synergies of arbitrary orders. We compared $\mathop {\mathtt {HAM}}\limits$ models with the most recent, state-of-the-art methods on six public benchmark datasets in three different experimental settings. Our experimental results demonstrate that $\mathop {\mathtt {HAM}}\limits$ models significantly outperform the state of the art in all the experimental settings. with an improvement as much as 46.6 percent. In addition, our run-time performance comparison in testing demonstrates that $\mathop {\mathtt {HAM}}\limits$ models are much more efficient than the state-of-the-art methods. and are able to achieve significant speedup as much as 139.7 folds.

关键词： Adaptation models Benchmark testing Trajectory Tools Recurrent neural networks Markov processes Image color analysis

Enhanced algorithmic modelling and architecture in deep reinforcement learning based on wireless communication Fintech technology

学校读者我要写书评

暂无评论

Optik 2023年 272卷

作者： Upreti, Kamal Syed, Mohammad Haider Khan, Mohiuddin Ali Fatima, Huda Alam, Mohammad Shabbir Sharma, A.K. Department of Computer Science & Engineering Dr. Akhilesh Das Gupta Institute of Technology & Management Delhi India College of Computing and Informatics Saudi Electronic University Saudi Arabia Department of Computer & Network Engineering Jazan University Jazan Saudi Arabia Department of Information Technology & Security Jazan University Jazan Saudi Arabia Department of Computer science College of Computer Science and Information Technology Jazan University Jizan Saudi Arabia Department of Creative Technologies & Product Design National Taipei University of Business Taipei City Taiwan

Financial technology" or "FinTech" refers to use of information technologies to derive financial solutions. FinTech is now widely regarded as a hotly debated blend of financial services and information technology. A combination of AI and IoT approaches will significantly increase the extraction of valuable financial data as well as provide better services to customers. This research proposed a novel protocol in data transmission using fuzzy rule-based secure transmission with data optimization technique using deep reinforcement neural network by wireless communication. This technique for data transmission will reduce the credit risk and enhances the data optimization by fuzzy rule-based protocol with DRNN. The logic utilized is fuzzy logic, which is a multi-valued logic with truth values for variables ranging from 0 to 1. When truth values range from entirely true to completely false, it is used. The end user's searching experience is improved by fuzzy logic-based semantic search, which finds and retrieves exact matching files for relevant search files provided by user. When exact matches aren't possible, method leverages semantic similarities to determine most relevant matches. A grading method is utilized to minimize number of false positives. For this, proposed technique was used, which may decide on a collection of suitable storage servers on which the data must be saved and resulting in a reduction in execution time while maintaining a higher level of security. The experimental results show the execution time of 54 %, network performance of 96 %, the overall complexity of 71 %, data optimization of 95 % and end-end delay of 67 %. © 2022

关键词： Semantics

Federated Learning for computer Vision

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Himeur, Yassine Varlamis, Iraklis Kheddar, Hamza Amira, Abbes Atalla, Shadi Singh, Yashbir Bensaali, Faycal Mansoor, Wathiq College of Engineering and Information Technology University of Dubai Dubai United Arab Emirates Department of Informatics and Telematics Harokopio University of Athens Greece LSEA Laboratory Electrical Engineering Department University of Medea Algeria Department of Computer Science University of Sharjah United Arab Emirates Department of Radiology Mayo Clinic RochesterMN United States Electrical Engineering Department Qatar University Qatar

computer Vision (CV) is playing a significant role in transforming society by utilizing machine learning (ML) tools for a wide range of tasks. However, the need for large-scale datasets to train ML models creates challenges for centralized ML algorithms. The massive computation loads required for processing and the potential privacy risks associated with storing and processing data on central cloud servers put these algorithms under severe strain. To address these issues, federated learning (FL) has emerged as a promising solution, allowing privacy preservation by training models locally and exchanging them to improve overall performance. Additionally, the computational load is distributed across multiple clients, reducing the burden on central servers. This paper presents, to the best of the authors’ knowledge, the first review discussing recent advancements of FL in CV applications, comparing them to conventional centralized training paradigms. It provides an overview of current FL applications in various CV tasks, emphasizing the advantages of FL and the challenges of implementing it in CV. To facilitate this, the paper proposes a taxonomy of FL techniques in CV, outlining their applications and security threats. It also discusses privacy concerns related to implementing blockchain in FL schemes for CV tasks and summarizes existing privacy preservation methods. Moving on, the paper identifies open research challenges and potential future research directions to further exploit the potential of FL and blockchain in CV applications. © 2023, CC BY.

关键词： computer vision

REDUCING OVERTREATMENT OF INDETERMINATE THYROID NODULES USING A MULTIMODAL DEEP LEARNING MODEL

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Athreya, Shreeram Melehy, Andrew Suthahar, Sujit Silas Armstrong Ivezić, Vedrana Radhachandran, Ashwath Sant, Vivek Moleta, Chace Zheng, Henry Patel, Maitraya Masamed, Rinat Arnold, Corey W. Speier, William Department of Electrical and Computer Engineering UCLA United States Department of Surgery UCLA United States Department of Bioengineering UCLA United States Medical Informatics UCLA United States Department of Surgery UT Southwestern Medical Center DallasTX United States Department of Pathology and Laboratory Medicine UCLA United States Department of Radiological Sciences UCLA United States

Objective: Molecular testing (MT) classifies cytologically indeterminate thyroid nodules as benign or malignant with high sensitivity but low positive predictive value (PPV), only using molecular profiles, ignoring ultrasound (US) imaging and biopsy. We address this limitation by applying attention multiple instance learning (AMIL) to US images. Methods: We retrospectively reviewed 333 patients with indeterminate thyroid nodules at UCLA medical center (259 benign, 74 malignant). A multi-modal deep learning AMIL model was developed, combining US images and MT to classify the nodules as benign or malignant and enhance the malignancy risk stratification of MT. Results: The final AMIL model matched MT sensitivity (0.946) while significantly improving PPV (0.477 vs 0.448 for MT alone), indicating fewer false positives while maintaining high sensitivity. Conclusion: Our approach reduces false positives compared to MT while maintaining the same ability to identify positive cases, potentially reducing unnecessary benign thyroid resections in patients with indeterminate nodules. © 2024, CC BY.

关键词： Contrastive Learning

Clinical-grade Multi-Organ Pathology Report Generation for Multi-scale Whole Slide Images via a Semantically Guided Medical Text Foundation Model

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Tan, Jing Wei Kim, SeungKyu Kim, Eunsu Lee, Sung Hak Ahn, Sangjeong Jeong, Won-Ki Department of Computer Science and Engineering College of Informatics Korea University Korea Republic of Department of Hospital Pathology Seoul St. Mary’s Hospital College of Medicine The Catholic University of Korea Korea Republic of Department of Pathology Korea University Anam Hospital College of Medicine Korea University Korea Republic of

Vision language models (VLM) have achieved success in both natural language comprehension and image recognition tasks. However, their use in pathology report generation for whole slide images (WSIs) is still limited due to the huge size of multi-scale WSIs and the high cost of WSI annotation. Moreover, in most of the existing research on pathology report generation, sufficient validation regarding clinical efficacy has not been conducted. Herein, we propose a novel Patient-level Multi-organ Pathology Report Generation (PMPRG) model, which utilizes the multi-scale WSI features from our proposed multi-scale regional vision transformer (MR-ViT) model and their real pathology reports to guide VLM training for accurate pathology report generation. The model then automatically generates a report based on the provided key features-attended regional features. We assessed our model using a WSI dataset consisting of multiple organs, including the colon and kidney. Our model achieved a METEOR score of 0.68, demonstrating the effectiveness of our approach. This model allows pathologists to efficiently generate pathology reports for patients, regardless of the number of WSIs involved. The code is available at https://***/hvcl/ Clinical-grade-Pathology-Report-Generation/tree/main © 2024, CC BY-NC-SA.

关键词： Visual languages

KappaFace: Adaptive Additive Angular Margin Loss for Deep Face Recognition

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Oinar, Chingis Binh, M.L.E. Woo, Simon S. Mercari Tokyo106-6118 Japan Department of Computer Science and Engineering College of Computing and Informatics Sungkyunkwan University Suwon16419 Korea Republic of

Feature learning is a widely used method for large-scale face recognition tasks. Recently, large-margin softmax loss methods have demonstrated significant improvements in deep face recognition. However, these methods typically propose fixed positive margins to enforce intra-class compactness and inter-class diversity, without considering imbalanced learning issues that arise due to different learning difficulties or the number of training samples available in each class. This overlook not only compromises the efficiency of the learning process but, more critically, the generalization capability of the resultant models. To address this problem, we introduce a novel adaptive strategy called KappaFace, which modulates the relative importance of each class based on its learning difficulty and imbalance. Drawing inspiration from the von Mises-Fisher distribution, KappaFace increases the margin values for the challenging or underrepresented classes and decreases that of more well-represented classes. Comprehensive experiments across eight cutting-edge baselines and nine well-established facial benchmark datasets strongly confirm the advantage of our method. Notably, we observed an enhancement of up to 0.5% on the verification task when evaluated on the IJB-B/C datasets. In conclusion, KappaFace offers a novel solution that effectively tackles imbalanced learning in deep face recognition tasks and establishes a new baseline. Copyright © 2022, The Authors. All rights reserved.

关键词： Face recognition

Active Neighbor Exploitation for Fast Data Aggregation in IoT Sensor Networks

学校读者我要写书评

暂无评论

TechRxiv

TechRxiv 2023年

作者： Vo, Van-Vi Le, Duc-Tai Raza, Syed M. Kim, Moonseong Choo, Hyunseung The Department of Electrical and Computer Engineering Sungkyunkwan University Suwon16419 Korea Republic of The College of Computing and Informatics Sungkyunkwan University Suwon16419 Korea Republic of The Department of IT Convergence Software Seoul Theological University Bucheon14754 Korea Republic of

Fast data aggregation is crucial to facilitate critical Internet of Things (IoT) services as it collects all sensory data under restricted volume and time using in-network computation. The minimum latency data aggregation scheduling problem has attracted a lot of attention for the last decade. Existing approaches for the problem schedule all data transmissions based on an aggregation tree which was typically constructed without secondary interference awareness. Such interference causes numerous delay when scheduling a transmission from a node to its parent in the tree. When a sensor is active at multiple timeslots in a working period, the secondary interference can simply be omitted by scheduling a node to transmit its data to an active neighbor instead of its parent. This paper presents an Active Neighbor EXploitation (ANEX) approach which enables sensor nodes to switch their parents by discovering active neighbors for possible connectivity, regardless of the receivers established in the tree. Furthermore, the scheme prioritizes scheduling nodes with the fewest unscheduled active neighbors in order to execute more concurrent transmissions. We evaluate ANEX by theoretical analysis and extensive simulations with various scenarios. The results show that ANEX outstandingly achieves faster aggregation than the state-of-the-art approach by up to 86% while having an equivalent time complexity. © 2023, CC BY.

关键词： Internet of things