检索结果-内蒙古大学图书馆

A novel approach for diabetic foot diagnosis: deep learning-based detection of lower extremity arterial stenosis

DIABETES RESEARCH AND CLINICAL PRACTICE 2024年 207卷 111032-111032页

作者： Wu, Chongxin Xu, Changpeng Ou, Shuanji Wu, Xiaodong Guo, Jing Qi, Yong Cai, Shuting Guangdong Univ Technol Sch Automat Guangzhou 510006 Peoples R China Guangdong Second Prov Gen Hosp Dept Orthopaed Guangzhou 510317 Peoples R China Guangdong Second Prov Gen Hosp Guangdong Tradit Med & Sports Injury Rehabil Res I Guangzhou 510317 Peoples R China Guangdong Univ Technol Sch Integrated Circuits Guangzhou 510006 Peoples R China

Purpose of the study: Assessing the lower extremity arterial stenosis scores (LEASS) in patients with diabetic foot ulcer (DFU) is a challenging task that requires considerable time and efforts from physicians, and it may yield varying results. The presence of vascular wall calcification and other irrelevant tissue information surrounding the vessel can further compound the difficulties of this evaluation. Automatic detection of lower extremity arterial stenosis (LEAS) is expected to help doctors develop treatment plans for patients ***: In this paper, we first reconstructed the 3D model of blood vessels by medical digital image processing and then utilized it as the training data for deep learning (DL) in conjunction with the non-calcified part of blood vessels in the original data. We proposed an improved model of vascular stenosis small target detection based on YOLOv5. We added Convolutional Block Attention Module (CBAM) in backbone, replaced Path Aggregation Network (PANET) with Bidirectional Feature Pyramid Network (BiFPN) and replaced C3 with GhostC3 in neck to improve the recognition of three types of stenosis targets (I: <50 %, II: 51 % -99 %, III: completely occluded). Additionally, we utilized K-Means++ instead of K-Means for better algorithm convergence performance, and enhanced the Complete-IoU (CIoU) loss function to Alpha-Scylla-IoU (ASIoU) loss for faster reasoning and convergence. Lastly, we conducted comparisons between our approach and five other prominent ***: Our method had the best average ability to detect three types of stenosis with 85.40% mean Average Precision (mAP) and 74.60 Frames Per Second (FPS) and explored the possibility of applying DL to the detection of LEAS in diabetic foot. The code is available at ***/wuchongxin/yolov5_***.

关键词： Diabetic foot disease Lower extremity arterial stenosis Automatic detection deep learning

来源：评论

学校读者我要写书评

暂无评论

First-in-human real-time AI-assisted instrument deocclusion during augmented reality robotic surgery

引用

HEALTHCARE TECHNOLOGY LETTERS 2024年第2-3期11卷 33-39页

作者： Hofman, Jasper De Backer, Pieter Manghi, Ilaria Simoens, Jente De Groote, Ruben Van Den Bossche, Hannes D'Hondt, Mathieu Oosterlinck, Tim Lippens, Julie Van Praet, Charles Ferraguti, Federica Debbaut, Charlotte Li, Zhijin Kutter, Oliver Mottrie, Alex Decaestecker, Karel ORSI Acad Melle Belgium Univ Ghent Fac Med & Hlth Sci Dept Human Struct & Repair Ghent Belgium Univ Ghent IBiTech Biommeda Fac Engn & Architecture Ghent Belgium Univ Ghent CRIG Ghent Belgium Ghent Univ Hosp Dept Urol Ghent Belgium Univ Modena & Reggio Emilia Dept Sci & Methods Engn Modena Italy Onze Lieve Vrouw Hosp Dept Urol Aalst Belgium AZ West Hosp Dept Urol Veurne Belgium AZ Groeninge Hosp Dept Digest & Hepatobiliary Pancreat Surg Kortrijk Belgium Katholieke Univ Leuven Fac Med Leuven Belgium NVIDIA Santa Clara CA USA ORSI Acad Proefhoevestr12 B-9090 Melle Belgium

The integration of Augmented reality (AR) into daily surgical practice is withheld by the correct registration of pre-operative data. This includes intelligent 3D model superposition whilst simultaneously handling real and virtual occlusions caused by the AR overlay. Occlusions can negatively impact surgical safety and as such deteriorate rather than improve surgical care. Robotic surgery is particularly suited to tackle these integration challenges in a stepwise approach as the robotic console allows for different inputs to be displayed in parallel to the surgeon. Nevertheless, real-time de-occlusion requires extensive computational resources which further complicates clinical integration. This work tackles the problem of instrument occlusion and presents, to the authors' best knowledge, the first-in-human on edge deployment of a real-time binary segmentation pipeline during three robot-assisted surgeries: partial nephrectomy, migrated endovascular stent removal, and liver metastasectomy. To this end, a state-of-the-art real-time segmentation and 3D model pipeline was implemented and presented to the surgeon during live surgery. The pipeline allows real-time binary segmentation of 37 non-organic surgical items, which are never occluded during AR. The application features real-time manual 3D model manipulation for correct soft tissue alignment. The proposed pipeline can contribute towards surgical safety, ergonomics, and acceptance of AR in minimally invasive surgery. This works presents the first-in-human edge deployment of a real-time AI-enabled augmented reality (AR) pipeline in robotic surgery. The application uses a binary segmentation model to effectively identify over 37 classes of non-organic items in the surgical scene, and uses this information to create an overlay visualization, solving the instrument occlusion problem, and preventing the possibly hazardous situation this implies, as well as adding a sense of depth to the AR. The solution is used during t

关键词： augmented reality computer vision image processing image segmentation learning (artificial intelligence) medical robotics real-time systems surgery

来源：评论

学校读者我要写书评

暂无评论

A Convolutional Neural Network-Based Method for real- time Eye State Identification in Driver Drowsiness Detection 3

A Convolutional Neural Network-Based Method for Real- Time E...

引用

3rd International Conference on Smart Generation Computing, Communication and Networking, SMART GENCON 2023

作者： Gill, Kanwarpartap Singh Anand, Vatsala Chauhan, Rahul Thapliyal, Siddhant Gupta, Rupesh Chitkara University Institute of Engineering and Technology Chitkara University Punjab India Graphic Era Hill University Computer Science and Engineering Uttarakhand Dehradun India Graphic Era Deemed to Be University Computer Science & Engineering Uttarakhand Dehradun India

ISBN: (纸本)9798350319125

The use of Convolutional Neural Networks (CNNs) for drowsiness detection is an AI-based method for detecting and warning of human drowsiness or weariness. deep learning models like convolutional neural networks (CNNs) learn by analysing images or videos, often focusing on human faces or eyes. Drowsiness cues may be recognised by these models, including drooping eyelids, yawning, and altered facial expressions. CNN-based sleepiness detection may be used to generate alarms or cautions using economic resources in real-time applications, such as driver monitoring systems, to help avoid accidents and improve safety. Artificial intelligence and computer vision access are at the heart of this technology, which aims to alleviate major safety risks associated with sleepiness and exhaustion. In this research, drowsiness detection and classification are accomplished with the help of a Sequential Convolutional Neural Network (CNN) model. The graphics processing unit (GPU) is used for producing preliminary data processing. The deep learning model is then shown graphically by analysing the Loss and Accuracy curves. With a predicted accuracy of 96%, the suggested model would pave the way for more research into sleepiness detection and categorization. © 2023 IEEE.

关键词： Artificial Intelligence Classification CNN Computer Vision deep learning deep learning Drowsiness Detection image processing Model Training Remote Sensing Sequential Model

来源：评论

学校读者我要写书评

暂无评论

Smart Weed Management and their Profile Mapping using deep learning and Digital Mapping 2

Smart Weed Management and their Profile Mapping using Deep L...

引用

2nd International Conference on Advancement in Computation and Computer Technologies, InCACCT 2024

作者： Tiwari, Amit Kumar Singh, Digvijay Yadav, Avanish Mishra, Saurabh Singh, Saloni United Institute of Technology Computer Science & Engineering Prayagraj India

ISBN: (纸本)9798350371314

Removal of weeds has been a challenging task for farmers. Conventional method of weed removal includes indiscriminate spraying of herbicides despite the fact that the presence of the weed is patchy. To find a better solution for weed management, two prototypes were developed using low-cost hardware and open-source deep learning models: Prototype-1, used a Raspberry Pi 4 B and YOLO v5 model while Prototype-2, used a more powerful NVIDIA GTX-1080 Ti GPU and YOLO v7 model for weed detection. Both prototypes were evaluated in custom potato fields, targeting common weeds like Chenopodium album and Portulaca oleracea while avoiding potato plants. The prototype-1, due to its low processing capabilities showed a significant lag in real-time detection of weed while the prototype-2 showed promising results with average precision of 97% and recall of 91%. A GPS device and Digital Mapping system was also integrated to calculate the effectiveness of Herbicide after evaluating the before and after results with the help of Smart Agro-Sprayer. These experiments show a promising future for the fusion of latest technologies like AI (deep learning), Automation, Digital Mapping and Data analytics to solve conventional agriculture problems and promote precision farming to increase crop yield with sustainable model. © 2024 IEEE.

关键词： Herbicides

来源：评论

学校读者我要写书评

暂无评论

Novel Framework for image Classification Based on Patch-Based CNN Model

Novel Framework for Image Classification Based on Patch-Base...

引用

International Conference on Data Analytics and Management, ICDAM 2023

作者： Gour, Ayush Bhanodia, Praveen Kumar Sethi, Kamal K. Rajput, Shivashankar Acropolis Institute of Technology and Research Indore Indore India

ISBN: (纸本)9789819965465

image data is being developed at a rapid speed every day due to the speedy growth of Internet;resultantly, image classification has become very crucial topic in solving computer vision-related tasks like images segmentation and object detection, etc. The convolutional neural networks (CNN) have become the preferred choice for categorizing images since 2012, because of its capability to find out hierarchical representations from the original image data without human intervention. CNNs are excellent at deciphering the intricate visual features present in images because they can capture local patterns and gradually build up more abstract and high-level representations. In several image processing tasks, patch-based solutions performed better than rival approaches. This research work illustrates the patch-based CNN model to increase picture classification accuracy, which may also serves as a model assembly without supplementary cost of model training. At the same time, applying data augmentation techniques throughout the training as well as testing phases helps assure network optimization as well as improvise generalization of the model. This methodology can transmit specific domain information effectively by processing individual patches instead of the entire image. To increase training samples, essential patches of appropriate size were extracted from the original images, improving overall model performance. Additionally, our approach is applicable to various network architectures and data augmentation techniques, making it versatile for deep learning tasks. Thorough evaluations on a real-world dataset demonstrate the superior performance of our proposed framework which also achieves an accuracy of 96.21% on the CIFAR-10 dataset, surpassing the state-of-the-art results (93.41%). © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2024.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Dynamic Online Modulation Recognition using Incremental learning

Dynamic Online Modulation Recognition using Incremental Lear...

引用

International Conference on Computing, Networking and Communications (ICNC)

作者： Owfi, Ali Abbasi, Ali Afghah, Fatemeh Ashdown, Jonathan Turck, Kurt Clemson Univ Holcombe Dept Elect & Comp Engn Clemson SC 29634 USA Vanderbilt Univ Dept Comp Sci Nashville TN 37235 USA Air Force Res Lab Rome NY USA

ISBN: (纸本)9798350371000;9798350370997

Modulation recognition is a fundamental task in communication systems as the accurate identification of modulation schemes is essential for reliable signal processing, interference mitigation for coexistent communication technologies, and network optimization. Incorporating deep learning (DL) models into modulation recognition has demonstrated promising results in various scenarios. However, conventional DL models often fall short in online dynamic contexts, particularly in class incremental scenarios where new modulation schemes are encountered during online deployment. Retraining these models on all previously seen modulation schemes is not only time-consuming but may also not be feasible due to storage limitations. On the other hand, training solely on new modulation schemes often results in catastrophic forgetting of previously learned classes. This issue renders DL-based modulation recognition models inapplicable in real-world scenarios because the dynamic nature of communication systems necessitate the effective adaptability to new modulation schemes. This paper addresses this challenge by evaluating the performance of multiple Incremental learning (IL) algorithms in dynamic modulation recognition scenarios, comparing them against conventional DL-based modulation recognition. Our results demonstrate that modulation recognition frameworks based on IL effectively prevent catastrophic forgetting, enabling models to perform robustly in dynamic scenarios.

关键词： Modulation Recognition Incremental learning deep learning Online learning

来源：评论

学校读者我要写书评

暂无评论

Advancing IR-UWB Radar Human Activity Recognition With Swin Transformers and Supervised Contrastive learning

引用

IEEE INTERNET OF THINGS JOURNAL 2024年第7期11卷 11750-11766页

作者： Li, Xiaoxiong Chen, Si Zhang, Shuning Zhu, Yuying Xiao, Zelong Wang, Xun Nanjing Univ Sci & Technol Sch Elect & Opt Engn Nanjing 210094 Peoples R China

Impulse radio ultrawideband (IR-UWB) radar has the advantages of low cost, high resolution, and independence of light and weather conditions. Its potential in human activity recognition (HAR) for IoT device sensing draws interest. One challenge in this domain is effectively representing spatial static and temporal dynamic information in echo sequences. Transformers, used extensively in NLP and CV, have powerful sequence long-range dependency modeling capabilities. However, in the field of radar HAR, the application research of transformers is still insufficient. In addition, there is currently a lack of publicly available IR-UWB radar human action data sets. To this end, we proposed various fine-grained feature image calculation methods and designed an IR-UWB Radar Human Activity data set (IURHA2023). This article presents a swin transformer encoder combining cosine similarity attention and patch overlap to obtain deep spatio-temporal features of human action feature images. Compared with other proposed transformer models or traditional CNNs and RNNs, the improved swin transformer encoder performs better. To further improve the feature learning capability of the backbone network and the robustness to echo variations, we propose a supervised contrastive learning-enhanced swin transformer (SCL-SwinT). It obtains distinctions and compact embeddings by comparing the similarities of positive and negative examples partitioned according to labels. Experimental results on the IURHA2023 data set show that SCL-SwinT achieves a recognition rate exceeding 90%, and the inference speed on IoT edge devices satisfies real-time applications. Ablation experiments demonstrate the effectiveness of the proposed components. In addition, SCL-SwinT exhibits good robustness to environmental factors like noise, multipath, and distance.

关键词： Radar Transformers Feature extraction Scattering Human activity recognition Adaptation models Ultra wideband radar Contrastive learning human activity recognition (HAR) Internet of Things impulse radio ultrawideband (IR-UWB) radar person identification transformer architecture

来源：评论

学校读者我要写书评

暂无评论

real time Spatial Sound Scene Analysis-AlertNet 3

Real Time Spatial Sound Scene Analysis-AlertNet

引用

International Conference on Advances in Computing, Communication and Applied Informatics (ACCAI)

作者： Nagaraj, S. Malathy, R. Sheeba, Merlin G. Jerusalem Engn Coll Chennai Tamil Nadu India

ISBN: (纸本)9798350389432;9798350389449;9798350389456

This project introduces a groundbreaking approach to enhance the accuracy of acoustic scene classification and ensures user safety within music player applications. It provides a combined model of deep learning by merging Convolutional Neural Networks (CNN) with Bidirectional Long Short-Term Memory (Bi-LSTM).The model uses Mel-Frequency Cepstral Coefficients (MFCC) and Gammatone Frequency Cepstral Coefficients (GFCC) features as feature extractors, and processes these features to predict ***, this paper integrates the advanced scene classification model into music player applications. Users can enjoy their favourite music while the system continuously analyses the audio environment. When a potentially dangerous situation is detected, such as sirens, alarms, or other alerting sounds, the music playback is momentarily attenuated or paused, and users are promptly alerted to pay attention to their surroundings. Performance evaluation demonstrates 97% accuracy in a noisy environment and 98 % accuracy in clean *** CNN-Bi-LSTM hybrid model, named AlertNet, excels in capturing both fine-grained and contextual audio information. Its seamless integration into music players enhances safety and user experience. Finally, it exemplifies the fusion of cutting-edge deep learning techniques with everyday technology to ensure safety and situational awareness for users, especially in scenarios where distractions are common.

关键词： Audio feature extraction MFCC,GFCC acoustic signal processing CNN Bi-LSTM Alerting system

来源：评论

学校读者我要写书评

暂无评论

Building Empathic Chabot using LSTM and NLP 4

Building Empathic Chabot using LSTM and NLP

引用

4th International Conference on Innovative Practices in Technology and Management, ICIPTM 2024

作者： Jeyalakshmi, Jeyalakshmi Srinidhi, S. Balaprathap, C. Department of Computer Science and Engineering Amrita VishwaVidhyapeetham Chennai India

ISBN: (数字)9798350307757

ISBN: (纸本)9798350307757

This research work presents a Empathic chatbot leveraging advanced natural language processing techniques. The chatbot employs a deep learning architecture, specifically a variant of the Transformer model, to generate human-like responses in real- time. The training dataset includes diverse conversational patterns to enhance the bot's ability to handle a wide range of user inputs. The model is fine-tuned through an iterative process, optimizing for coherence, relevance, and contextual understanding. Evaluation metrics demonstrate the effectiveness of the generative chatbot in providing contextually appropriate and coherent responses across various domains. The research contributes to the evolving field of conversational AI, showcasing the potential of generative models for creating engaging and context-aware chatbot interactions. © 2024 IEEE.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Advancing 3D Mesh Segmentation with Regularization Losses: A Semi-Supervised Approach Using MeshCNN

Advancing 3D Mesh Segmentation with Regularization Losses: A...

引用

2024 International Symposium on Digital Home, ISDH 2024

作者： Xing, Yan Lv, Shuaijun Li, Daolun Zhuang, Wanliang Tan, Jieqing Lan, Rushi School of Mathematics Hefei University of Technology Hefei China Guilin University of Electronic Technology Guangxi Key Lab of Image and Graphic Intelligent Processing Guilin China

ISBN: (纸本)9798331509873

The application of deep learning in 3D mesh segmentation has attracted increasing attention. In supervisedlearning based mesh segmentation, back propagation is often based on cross entropy loss. Some prior knowledge, such as the structure similarity or geometric consistency of data, can be used to improve the segmentation effect. However, in the literature for mesh segmentation, such prior knowledge is usually either ignored in deep learning or only considered at post-processing stage. In this work, we add prior knowledge as a regularization term into the loss function for optimization, which is applied in the semi-supervised 3D mesh segmentation based on MeshCNN for the first time. A large number of experiments show that, for both semi-supervised segmentation with 30% and 70% labeled data and fully supervised segmentation, the proposed regularization term improves the segmentation performance. © 2024 IEEE.

关键词： Mesh generation

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：