检索结果-内蒙古大学图书馆

Neural network-based small cursor detection for embedded assistive technology

Visual computer 2024年第12期40卷 8425-8439页

作者： Darmawan, Jeremie Theddy Sigalingging, Xanno Kharis Faisal, Muhamad Leu, Jenq-Shiou Ratnasari, Nanda Rizqia Pradana Department of Electronic and Computer Engineering National Taiwan University of Science and Technology No. 43 Section 4 Keelung Rd Da’an District Taipei City106 Taiwan Department of Bioinformatics School of Life Sciences Indonesia International Institute for Life Sciences Jl. Pulomas Barat No.Kav. 88 RT.4/RW.9 Kayu Putih Kec. Pulo GadungSpecial Capital Region of Jakarta East Jakarta13210 Indonesia Department of Biochemistry Yong Loo Lin School of Medicine National University of Singapore C/O MD7 Level 2 8 Medical Drive Singapore117597 Singapore

Assistive technology (AT) is invaluable to people with special educational needs and disabilities, enabling them to interact with computers more efficiently. It is able to improve the interactivity between humans and computers for learning purposes. However, there is a lack of accessibility and interactivity with touchscreen devices and visual content that AT users encounter. These issues are critical in providing much-needed assistance and enhancing the daily lives of disabled individuals. Therefore, we propose an AT framework based on neural networks using embedded systems for improving the accessibility and interactivity of AT users. The proposed framework utilizes the Jetson Nano and is mainly used with a speech-to-intent neural network to process speech and move cursors. When improved with cursor object detection, the framework could obtain the location of the cursors in external displays and move the cursors of other devices. Since cursor datasets are very limited and not many detection models are up for the task, we investigated the use of Slicing Aided Hyper Inference (SAHI) pipeline along with two fine-tuned models, Fully Convoluted One-Stage (FCOS) and Task-aligned One-stage Object Detection (TOOD), to identify the minimum data required for these models to work optimally. With less than 120 annotated images and a data multiplier of 5 and 30, both models were able to achieve ~ 52 and ~ 60mAP, respectively. These results were comparable to performance on other small object detection datasets. In addition, we also present a working proof-of-concept for our proposed embedded assistive technology framework. © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.

关键词： Assistive technology

来源：评论

学校读者我要写书评

暂无评论

Comprehensive analysis on Ultralytics-supported YOLO models for detection and recognition of large office objects for indoor navigation

引用

Procedia computer science 2024年 246卷 3851-3858页

作者： Renann G. Baldovino Aron Jake P. Vidad Rudwin Paul B. Abastillas Nilo T. Bugtai Elmer P. Dadios Ryan Rhay P. Vicerra Argel A. Bandala Aaron Raymond See Nicanor R. Roxas Institute of Biomedical Engineering and Health Technologies (IBEHT) Department of Manufacturing Engineering and Management (DMEM) Department of Electronics and Computer Engineering (DECE) De La Salle University (DLSU) 2401 Taft Avenue 0922 Manila Philippines Department of Electrical Engineering Southern Taiwan University of Science and Technology (STUST) No. 1號 Nantai St Yongkang District Tainan City 71005 Taiwan Department of Electronic Engineering National Chin-Yi University of Technology No. 57 Section 2 Zhongshan Rd Taiping District Taichung City 411030 Taiwan

The state of object detection has quickly progressed due to the rapid development of hardware and deep learning models for the said task. The continuous development of the latter allowed its application in real-time. That is, from two-stage to one-stage detectors, which greatly reduced the processing time; and from anchor-based to anchor-free detectors, which allowed a more efficient and flexible model implementation. The purpose of this study is to explore the various YOLO versions available in Ultralytics to detect large office objects for obstacle avoidance. More specifically, this study observed how the versions, 3u, 5mu, 8 and 6m, from which the first three are anchor-free, perform over the dataset. Based from the results, it was found that there were no significant differences between the detectors, except for the anchor-based v6m, which has the worst scores. The four models, v8n, v8s, v8m, and v5mu, were, however, chosen as the candidate models due to their processing speed, ranging from 66 to 75 FPS on the NVIDIA GeForce RTX 3060 GPU. The said models were also subjected to an optimization process in terms of the confidence and IoU threshold values to find the pair that could provide the best metric scores. Results have provided candidate models for the study’s detection task for obstacle avoidance in real-time.

关键词： deep learning indoor navigation object recognition YOLO

来源：评论

学校读者我要写书评

暂无评论

Photovoltaic Induced Self-Powered Gas Sensor Based on 2d Mos2 Incorporated Nbse2 Nanorods Heterostructure for Nh3 Gas Sensing at Room Temperature

SSRN

引用

SSRN 2024年

作者： Saravanan, Adhimoorthy Huang, Bohr-Ran Hwang, Seung-Kyu Kathiravan, Deepa Hsiao, Wesley Wei-Wen Ravichandran, Chitra Abun, Abebaw Hong, Po-Da Mohammadi, Ali Vilian, Ezhil Han, Young-Kyu Huh, Yun Suk Graduate Institute of Electro-Optical Engineering Department of Electronic and Computer Engineering National Taiwan University of Science and Technology Taipei106 Taiwan Department of Materials Science and Engineering National Taiwan University of Science and Technology 43 Keelung road Section 4 Taipei10607 Taiwan Department of Chemical Engineering National Taiwan University of Science and Technology Taipei10607 Taiwan Department of Biological Sciences and Bioengineering Nano Bio High-Tech Materials Research Center Inha University Incheon22212 Korea Republic of Department of Energy and Materials Engineering Dongguk University-Seoul Seoul100-715 Korea Republic of

Two-dimensional (2D) transition metal dichalcogenides (TMDs) have attracted significant attention for their optical and gas-sensing applications due to their exceptional sensitivity. Reliable selectivity and low power consumption are two major requirements for photodetector and gas sensor applications in next-generation electronic devices and the Internet of Things. Self-powered sensors (especially photovoltaic gas sensors) can solve these problems. In this study, for the first time, we propose 2D TMDs (NbSe2-MoS2 hybrid) on a SiO2/Si substrate to fabricate photovoltaic self-powered gas sensors. The gas sensors are operated by the photovoltaic effect of the NbSe2-MoS2 nanostructure, which is prepared using the liquid phase exfoliation process. Initially, it was revealed that the present hybrid material exhibits photovoltaic properties under light illumination, with a circuit current of 0.25 µA and a circuit voltage of 34 mV. The NbSe2-MoS2 nanostructure characteristics were then used for NH3 gas sensing at different concentrations, and the gas sensing response was detected from low (8.8% at 10 ppm) to high (28.8% at 500 ppm) concentrations. The built-in electric field occurred between the NbSe2-MoS2 junction and eventually operated as a driving force for NbSe2-MoS2 gas sensing without an external bias voltage. The physisorption of gas molecules on their surface prompts a charge-transfer mechanism that improves the gas sensor response. The combined outcome of NbSe2-MoS2 heterostructures could pave way to next-generation gas sensing device fabrications. © 2024, The Authors. All rights reserved.

关键词： Molybdenum compounds

来源：评论

学校读者我要写书评

暂无评论

Development of Localized Cutaneous Force Feedback System for Robotics Assisted Surgery Systems 28th

Development of Localized Cutaneous Force Feedback System for...

引用

28th International Conference on Knowledge Based and Intelligent information and engineering Systems, KES 2024

作者： See, Aaron Raymond Tiong, Thad Jacob De Guzman, Lanz Benedict Contee, Kenneth Lim, Garfield Greg Yebes, Carl Sebastian Department of Electrical Engineering Southern Taiwan University of Science and Technology No. 1 Nantai St Yongkang District Tainan City71005 Taiwan Department of Electronic Engineering National Chin-Yi University of Technology No. 57 Section 2 Zhongshan Rd Taiping District Taichung City411030 Taiwan Department of Electrical Engineering Universitas Katolik Indonesia Atma Jaya Jl. Jend. Sudirman No.51 5 RT.004/RW.4 Karet Semanggi Kecamatan Setiabudi Kota Jakarta Selatan Daerah Khusus Jakarta12930 Indonesia School of Computer Studies University of San Jose - Recoletos Basak Pardo Cebu City6000 Philippines College of Computer Studies Cebu Institute of Technology - University Natalio B. Bacalso Ave Cebu City Cebu6000 Philippines

This study explores integrating a novel cutaneous force feedback system for robotic-assisted surgery (RAS) systems aimed at improving tactile perception for surgical precision and safety. The system developed comprises two major components: a force-sensitive matrix sensor and a 4x4 tactile pin interface. Forces experienced by the sensors are mirrored through the tactile interface, conveying the sensation of knowing which part of the sensor is in contact and how much force is subjected to it. An algorithm called Dynamic Interval Localization (DIL) was developed to manipulate pin actuation by scheduling individualized dynamic oscillation. The system successfully detected operating forces from 0.1N to 5N while concurrently providing respective cutaneous feedback. The system's performance in force localization suggests significant potential and promises in elevating the haptic features of RAS systems. © 2024 The Authors.

关键词： Robotic surgery

来源：评论

学校读者我要写书评

暂无评论

Green Leaf Disease Detection Using Raspberry Pi and Random Forest Machine Learning Algorithm: An Automated Approach for Precision Agriculture

Green Leaf Disease Detection Using Raspberry Pi and Random F...

引用

International Conference on Smart Computing & Communications (ICSCC)

作者： Chitra Sabapathy Ranganathan Amjath Ali Jakul Basha A Vijayalakshmi N. Mohankumar S. Sujatha C. Sasikala Mphasis Corporation Chennai Tamil Nadu India Department of Engineering Computer Engineering Electrical and Electronic Section University of Technology and Applied Sciences Ibra Sultanate of Oman Department of Computer Science and Engineering Chennai Institute of Technology Chennai Tamil Nadu India Symbiosis Institute of Technology Symbiosis International (Deemed University) Nagpur Maharashtra India Department of Biomedical Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Saveetha University Chennai Tamil Nadu India Department of Electronics and Communication Engineering Sri Sai Ranganathan Engineering College Coimbatore Tamil Nadu India

ISBN: (数字)9798350363104

ISBN: (纸本)9798350363111

Farm production and food security are affected by green leaf diseases. Modern disease detection technologies may discover plant health issues early. This work uses Raspberry Pi, a cheap and flexible single-board computer, along the advanced Random Forest machine learning technique to identify a green leaf disease. To identify crop diseases, the study uses automated, non-invasive technologies. From labeled images of good and sickly leaves, the algorithm extracts color histograms and texture, and form. It feeds a Random Forest classifier these properties for classifying data with high dimensions. Several green leaf diseased plant species were examined. This method detected plant health concerns with high precision. Movement and low cost make the Raspberry Pi ideal for remote and resource-limited agriculture, increasing farming accuracy. Helping farmers and users identify and develop green leaf diseases improves efficient farming. This methodology has the potential to improve crop management, optimize pesticide usage, and increase agricultural productivity. These results might contribute to the development of advanced automatic disease detection techniques to address green leaf diseases in modern agriculture.

关键词： Productivity Precision agriculture Machine learning algorithms Accuracy Plants (biology) Crops Real-time systems Random forests Diseases Farming

来源：评论

学校读者我要写书评

暂无评论

Performance Analysis of Machine Learning Algorithm for the Credit Risk Analysis in the Banking Sector

Performance Analysis of Machine Learning Algorithm for the C...

引用

International Conference on Computing Methodologies and Communication (ICCMC)

作者： Sandeep Kumar Hegde Rajalaxmi Hegde Krishna Priya R Sree Southry S A.V.G.A. Marthanda K. Logu Department of Computer Science and Engineering NMAM Institute of Technology Nitte (Deemed to be University) Karkala Taluk Udupi District India Department of Engineering Electrical Section University of Technology and Applied Sciences Shinas campus Oman Department of ECE Sona College of Technology Salem India Department of EEE LakiReddy BaliReddy college of Engineering Mylavaram Permanently affiliated to JNTU Kakinada Andhrapradesh India Department of Computer Science and Engineering Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai Tamilnadu India

The banking sector has advanced in recent years. Thus, there is an increase in the demand for bank loans. The bank must distribute and sell each of the limited number of available slots to a select group of people. As a result, a usual stage is to identify who will be unable to return the loan and who will prove to be a more trustworthy option to the bank. In order to save the bank time and costs, in the proposed paper machine learning based approach is introduced to reduce the risk involved with finding the safe individual. In order to decide whether or not to grant someone a loan, this paper presents a method of loan approval based on predetermined criteria. The machine learning model for credit approval was implemented using logistic regression, XG Boost, random forest and naïve bayes model. The experimental results indicates that logistic regression model is more accurate for the credit risk analysis.

关键词： Analytical models Machine learning algorithms Costs Computational modeling Banking Performance analysis Risk analysis

来源：评论

学校读者我要写书评

暂无评论

Rough Fermatean fuzzy decision-based approach for modelling IDS classifiers in the federated learning of IoMT applications

引用

Neural Computing and Applications 2023年第30期35卷 22531-22549页

作者： Albahri, O.S. Al-Samarraay, Mohammed S. AlSattar, H.A. Alamoodi, A.H. Zaidan, A.A. Albahri, A.S. Zaidan, B.B. Jasim, Ali Najm Computer Techniques Engineering Department Mazaya University College Thi-Qar Nassiriya12733 Iraq Faculty of Computing and Meta-Technology Universiti Pendidikan Sultan Idris Perak Tanjong Malim35900 Malaysia Department of Business Administration College of Administrative Science The University of Mashreq Baghdad10021 Iraq SP Jain School of Global Management Lidcombe SydneyNSW2141 Australia Iraqi Commission for Computers and Informatics Baghdad10022 Iraq Future Technology Research Center College of Future National Yunlin University of Science and Technology 123 University Road Section 3 Yunlin Douliou64002 Taiwan Foundation of Alshuhda Baghdad Iraq MEU Research Unit Middle East University Amman Jordan

Intrusion detection systems (IDSs) are commonly employed to mitigate network security threats in various fields, including federated learning applications within the Internet of Medical Things (IoMT). However, IDSs face challenges owing to the sheer volume of network traffic, high-dimensional datasets and the necessity for real-time detection. Although machine learning integration assists IDSs in overcoming these challenges, modelling difficulties persist due to varied evaluation criteria and levels of conflict and importance. Multi-criteria decision-making (MCDM) solutions have been utilised in IoMT and IDS, yet they fall short in capturing the subjective judgements of experts and rely on normalisation approaches, which can impact results. This study seeks to address these issues through the integration of robust MCDM methodologies, namely fuzzy-weighted zero-inconsistency (FWZIC) and fuzzy decision by opinion score method (FDOSM). Utilising rough Fermatean fuzzy sets (RFFSs), this integration produces precise solutions with reduced uncertainty. Our methodology involves adopting a decision matrix for IDS classifiers based on integrated evaluation criteria, followed by deriving new formulations and developments for RFFSs-based FDOSM and FWZIC for the modelling and weighting of criteria, respectively. Evaluations using datasets involving 125,973 records and 41 features across 17 evaluation criteria revealed that accuracy–security and training time–performance weights yielded the highest scores, whereas false negative rate–security and CPU time–performance criteria received the lowest weights. The random forest emerged as the optimal IDS classifier. Systematic modelling, sensitivity analysis and comparative studies confirmed the robustness of our results. © 2023, Crown.

关键词： Integration

来源：评论

学校读者我要写书评

暂无评论

Designing a Trajectory Control System for Wheelchairs with a Combination of Automatic and Manual Steering Methods

引用

Periodica Polytechnica Transportation engineering 2024年第4期52卷 412-422页

作者： Vu, Van Tan Nguyen, Quang Thinh Nakagawa, Chihiro Nguyen, Trung Nguyen Le, Thanh Long Nguyen, Van Vinh Hoang, Thi Lan Anh Do, Trong Tu Mihály, András Gáspár, Péter Department of Automotive Mechanical Engineering Faculty of Mechanical Engineering University of Transport and Communications Building A6 no. 3 Cau Giay Street Lang Thuong Ward Dong Da District Hanoi100000 Viet Nam Japan Technology Solution Joint Stock Company 9 Dau Ma Street Section 2 Thi Cau Ward Bac Ninh Bac Ninh222000 Viet Nam School of Engineering Osaka Metropolitan University 1-1 Gakuen-cho Nakaku Sakai Osaka599-8531 Japan Emergency abdominal surgery Department Viet Duc University Hospital no. 40 Trang Thi Street Hanoi100000 Viet Nam Mechanical and Power Engineering Faculty Electric Power University no. 235 Hoang Quoc Viet Street Co Nhue Ward Bac Tu Liem District Hanoi100000 Viet Nam Systems and Control Laboratory Institute for Computer Science and Control Hungarian Research Network Kende u. 13-17 Budapest1111 Hungary

A wheelchair is a specialized form of Personal Mobility Vehicle (PMV) designed to help people with disabilities move safely to their desired locations. Unlike conventional PMVs, wheelchairs for people with disabilities are challenging to manually control in environments with heavy traffic. This study focuses on designing and feasibility testing of control systems for PMVs to meet safety and efficiency standards, even in complex environments. First, a control structure model for PMVs is proposed. This is followed by the application of a combination of First In First Out (FIFO) and Dijkstra traffic control methods to determine effective trajectories for PMVs. The proposed control system is then tested on the campus of the University of Transport and Communications. Simulation results show that when comparing the travel time of 10 PMVs, the order execution time for each PMV remains stable (within 10%) for up to 20 PMVs. However, when the number of PMVs increases to 25, 30, 35, and 40 vehicles, the order execution time increases by 35%, 55%, 103%, and 119%, respectively. These results demonstrate that the system can control a large number of PMVs, ensuring safety, flexibility, and efficiency. © 2024 Budapest University of Technology and Economics. All rights reserved.

关键词： Travel time

来源：评论

学校读者我要写书评

暂无评论

Towards a Personalized LLM-Based Daily Edge Memory Aid

Towards a Personalized LLM-Based Daily Edge Memory Aid

引用

International Conference on Information and Communication Technology Convergence (ICTC)

作者： Sangil Ha Dongbeom Ko Joon-Ho Lim Sungjoo Kang Hyeon Soo Kim AI Lab Tutorus Labs Inc. Deajeon Republic of Korea AI Computing System SW Research Section Electronics and Telecommunications Research Institute (ETRI) Deajeon Republic of Korea Department of Computer Science and Engineering Chungnam National University (CNU) Deajeon Republic of Korea

ISBN: (数字)9798350364637

ISBN: (纸本)9798350364644

This paper introduces a personalized memory aid system that combines a high-performance cloud-based large language model (LLM) with a low-power edge device running a small language model (sLM) to enhance user productivity by processing and summarizing daily conversations in real-time. The system optimizes efficiency and user experience while ensuring privacy through localized data handling. By combining cloud and edge resources, the system provides scalable, real-time memory support without compromising data security. The research highlights the system's architecture, its role in protecting user data, and its potential to seamlessly integrate into daily life, offering a sustainable and efficient approach to personalized artificial intelligence applications.

关键词： Performance evaluation Productivity Data privacy Systems architecture Oral communication Data processing Real-time systems Hardware User experience Resource management

来源：评论

学校读者我要写书评

暂无评论

Complete Frequency-Bin Bell Basis Synthesizer

引用

Physical Review Letters 2022年第23期129卷 230505-230505页

作者： Suparna Seshadri Hsuan-Hao Lu Daniel E. Leaird Andrew M. Weiner Joseph M. Lukens Elmore Family School of Electrical and Computer Engineering and Purdue Quantum Science and Engineering Institute Purdue University West Lafayette Indiana 47907 USA Quantum Information Science Section Oak Ridge National Laboratory Oak Ridge Tennessee 37831 USA Research Technology Office Arizona State University Tempe Arizona 85287 USA

We report the experimental generation of all four frequency-bin Bell states in a single versatile setup via successive pumping of spontaneous parametric down-conversion with single and dual spectral lines. Our scheme utilizes intensity modulation to control the pump configuration and offers turn-key generation of any desired Bell state using only off-the-shelf telecommunication equipment. We employ Bayesian inference to reconstruct the density matrices of the generated Bell states, finding fidelities ≥97% for all cases. Additionally, we demonstrate the sensitivity of the frequency-bin Bell states to common-mode and differential-mode temporal delays traversed by the photons comprising the state—presenting the potential for either enhanced resolution or nonlocal sensing enabled by our complete Bell basis synthesizer.

关键词： Entanglement production Quantum correlations in quantum information Quantum entanglement Quantum optics Quantum state engineering

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：