检索结果-内蒙古大学图书馆

LeXNet++: Layer-wise eXplainable ResUNet++ framework for segmentation of colorectal polyp cancer images

Neural Computing and Applications 2025年第1期37卷 213-229页

作者： Das, Surajit Khan, Soumya Suvra Sengupta, Diganta Debashis, De Dept. of Information Technology Meghnad Saha Institute of Technology Behind Urbana Complex Near Ruby General Hospital Anandapur Rd Uchhepota Kolkata700150 India Dept. of Computer Science & Engineering Meghnad Saha Institute of Technology Behind Urbana Complex Near Ruby General Hospital Anandapur Rd Uchhepota Kolkata700150 India Dept. of Computer Science & Engineering Heritage Institute of Technology Chowbaga Road West Bengal Anandapur Kolkata700107 India Dept. of Computer Science & Engineering Maulana Abul Kalam Azad University of Technology West Bengal NH-12 Simhat Haringhata West Bengal Nadia741249 India

Colorectal polyps are benign lesions that develop in the colon and can progress to cancer if left untreated. Clinical observations from medical images are often preferred over computational results due to the lack of trust in the machine learning models, thereby posing serious challenge for the explainability of the results. In order to computationally diagnose colorectal polyps from cancerous images and explain the results, we propose a Layer-wise eXplainable ResUNet++ (LeXNet++) framework for segmentation of the cancerous images, followed by layer-wise explanation of the results. We utilize a publicly accessible dataset that contains of 612 raw images with a resolution of 256×256×3 and an additional 612 clinically annotated and labeled images with a resolution of 256×256×1, which includes the infected region. The LeXNet++ framework comprises of three components—encoder, decoder and the bridge. The encoder and the decoder components each comprise of four layers. Each of the four layers in the encoder and the decoder comprises of 14 and 11 internal sub-layers, respectively. Among the sub-layers of the encoder and the decoder, there are three 3×3 convolutional layers with an additional 3×3 convolution-transpose layer in the decoder. The output of each of the sub-layers has been explained through heatmap generation after each iteration which have been further explained. The encoder and the decoder are connected by the bridge which comprises of three sub-layers. The results obtained from these three sub-layers have also been explained to inculcate trust in the findings. In this study, we have used three models to segment the images, namely UNet, ResUNet, and proposed LeXNet++. LeXNet++ exhibited the best result among the three models in terms of performance;hence, only LeXNet++ was explained layer-wise. Apart from explanation of the results fetched in this study, the performance of the proposed explainable model has been observed to be 2% greater than the existing poly

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Design of a Hybrid Framework for Detecting Attacks in Fog-based IoT Architecture using Deep Learning 15

Design of a Hybrid Framework for Detecting Attacks in Fog-ba...

引用

15th International Conference on Advances in Computing, Control, and Telecommunication Technologies, ACT 2024

作者： Pokale, Navnath B. Sharma, Pooja Mane, Deepak T. Dept. of Computer Engineering School of Engineering and Technology D. Y. Patil University Ambi Pune410507 India Dept. of Computer Engineering Vishwakarma Institute of Technology Pune India

ISBN: (纸本)9798331300579

IoT-Fog computing offers a broad variety of services for IoT-based end-systems. End IoT devices communicate with cloud nodes and fog nodes to administer client tasks. During the data collection process between the fog and cloud layers, IoT endpoints are more likely to be compromised by critical attacks such as DDoS and other security threats. It is necessary to detect these network vulnerabilities early on. By extracting features and placing the danger in the network, DL is crucial in predicting end-user behavior. Due to the limited processing power and less storage, Deep Learning algorithms cannot be directly implemented on IoT devices. In this article, we present a three stage Deep Hybrid Detection Model for Attack Detection in IoT-Fog Architecture. Enhanced Z-score normalization based data preparation will be carried out in the beginning. On the basis of preprocessed data, features such as IG, raw data, entropy, and enhanced MI are extracted in the second step. The collected characteristics are used as an input to hybrid classifiers called DBN and optimized DeepMaxout in the third stage to classify the assaults based on the input dataset. A hybrid optimization model known as the BMJFO (Blue Monkey Jelly Fish Optimization) algorithm is presented for the best DeepMaxout training. The suggested model produced increased accuracy, precision, sensitivity, and specificity. The suggested model produced superior accuracy, precision, sensitivity, and specificity of 95.80%, 94.94%, 96.38%, and 97.94%, respectively. © Grenze Scientific Society, 2024.

关键词： Fog computing

来源：评论

学校读者我要写书评

暂无评论

Employing Convolutional Neural Network for Real-time Detection and Localization of Physical Contact in a Redundant Manipulator Body 12

Employing Convolutional Neural Network for Real-time Detecti...

引用

12th RSI International Conference on Robotics and Mechatronics, ICRoM 2024

作者： Jafari, Amirhossein Tafti, Ali Abolhasani Karami, Abbas Isfahan University of Technology Dept. of Electrical and Computer Engineering Isfahan Iran Isfahan University of Technology Dept. of Mechanical Engineering Isfahan Iran Shiraz University of Technology Dept. of Mechanical Engineering Shiraz Iran

ISBN: (纸本)9798331529734

In the developing field of human-robot collaboration, robots increasingly share their workspace with humans, which becomes necessary to achieve seamless and safe interactions for manual guidance of robots in physical cooperation scenarios. The present study proposes a new method based on a convolutional neural network (CNN) for safe human-robot cooperation in the presence of intentional physical interactions, which is able to detect the occurrence and localization of contact on the robot arm in real-time. In this approach, contact detection and its localization are done only using the online external torque vector obtained by the residual-based observer of the robot, which eliminates any need to use additional force and vision sensors. Unlike traditional methods that depend on predefined thresholds or model-based techniques, our presented architecture with one-dimensional CNN can automatically learn the patterns from collected raw torque data and, by analyzing them, can properly predict the contact position. The performance of the proposed approach during manipulation has been validated through multiple online experiments in static and dynamic states on a KUKA LBR iiwa robot. © 2024 IEEE.

关键词： Redundant manipulators

来源：评论

学校读者我要写书评

暂无评论

Cost-balanced Adaptive Replica Server Cache Management under Time-Varying User Request Density in CCDNs 14

Cost-balanced Adaptive Replica Server Cache Management under...

引用

14th IEEE International Conference on Communication Systems and Network Technologies, CSNT 2025

作者： Gupta, Rohit Kumar Mazumdar, Arka Prokash Dept. of Computer Science & Engineering Malaviya National Institute of Technology Jaipur Manipal University Jaipur India Dept. of Computer Science & Engineering Malaviya National Institute of Technology Jaipur Jaipur India

ISBN: (纸本)9798331531935

In today’s growing Internet, cost-effective on-demand provisioning of caching resources in Cloud-based Content Delivery Networks (CCDNs) is essential to preserve the cache hit ratio while reducing storage requirements. However, during the high and fluctuating request rates at different locations, the sudden increase in cache to mitigate the miss rate requires a balance between storage cost (from dynamic cache provisioning) and miss cost (from latency in retrieving content from the Origin Server (OS) or other Replica Servers (RSs)). In such an environment, some RS instances remain underutilized while others are overburdened across different locations and time intervals, which requires effective RS management to activate only sufficient RSs by adapting to the user demands. To overcome these challenges, a cost-balanced adaptive cache management approach is proposed for dynamic cache provisioning in CCDN. This method aims to dynamically manage the activation of RS to optimize resource utilization, and cache instance provisioning and content caching operations in these RSs are managed by introducing an Adaptive Cache Manager that retrieves content from the OS or other RSs by effectively balancing costs in different time intervals. The experimental results indicate that the proposed method reduces the number of active replica servers in CCDN across different time intervals while decreasing the average total cost by about 56% relative to the LRU caching policy on fixed cache and about 10% compared to the dynamic selective caching policy on RCW cache. © 2025 IEEE.

关键词： Cache memory

来源：评论

学校读者我要写书评

暂无评论

Efficiency Redefined: The Document Scanner App for Precise Entity Identification with Customized NER 3

Efficiency Redefined: The Document Scanner App for Precise E...

引用

3rd IEEE Delhi Section Flagship Conference, DELCON 2024

作者： Javid, Sajid Saim, Mohammad Islam, Saiful Indraprastha Institute of Information Technology Delhi Dept. of Computer Science & Engineering Delhi India University of Cincinnati Dept. of Computer Science & Engineering CincinnatiOH United States Zakir Husain College of Engineering & Technology Aligarh Muslim University Dept. of Computer Engineering Aligarh India

ISBN: (纸本)9798331518592

In the era of rapid advancements in technology, the efficient digitization of paper-based documents remains a crucial challenge across various domains. The traditional approach to document scanning often struggles with accurate extraction of critical information from unstructured or diverse document formats. This paper presents the development of a web-based document scanner application that leverages Named Entity Recognition (NER) to extract entities from scanned business cards efficiently. The study aims to address the challenge of digitizing and organizing physical documents, with a focus on precise entity extraction. The methodology involves creating a custom dataset of scanned business cards, preprocessing the data, and training an NER model using machine learning techniques. The application is built using the Flask framework and integrates OpenCV for text extraction and SpaCy for NER. The key findings include the model's high F1 score in identifying entities such as names, organizations, designations, emails, and contact details. The application streamlines document digitization by automatically extracting and tagging relevant metadata, minimizing the need for manual data entry. This research contributes to enhancing document management and demonstrates broader applicability for other document types apart from business cards. The implications highlight improvements in workflow automation, data accuracy, and the potential for integrating with larger document management systems. © 2024 IEEE.

关键词： Optical character recognition

来源：评论

学校读者我要写书评

暂无评论

Small Object Detection for UAVs Using Deep Learning Models on Edge Computing: A Comparative Analysis 5

Small Object Detection for UAVs Using Deep Learning Models o...

引用

5th International Conference on Circuits, Control, Communication and Computing, I4C 2024

作者： Vasishta, M S Saathvik Reddy Amireddy, Abhinava Teja Shrivastava, Pranjal Raghuram, S. Talasila, Viswanath Shastry, Prasad N Ramaiah Institute of Technology Dept. of Electronics and Telecommunication Engineering Bangalore India Ramaiah Institue of Technology Dept. of Electronics and Communication Engineering Bangalore India Ramaiah Institute of Technology Dept. of Computer Science Engineering Bangalore India Ramaiah Institute of Technology Dept. of Electronics and Communication Engineering Bangalore India Bradley University Dept. of Electrical and Computer Engineering PeoriaIL United States

ISBN: (纸本)9798331528539

This paper presents a comparative analysis of DNN-models, including TensorFlow Lite, SSD-MobileNet, Mask R-CNN and YOLOv8, for the high-precision detection and classification of small objects in aerial search and rescue scenarios. The study features a proprietary dataset collected through drone flights over an extended period. The model performance was evaluated on compact computing platforms like Jetson Nano and Raspberry Pi, emphasizing detection accuracy and computational efficiency. Results reveal trade-offs between model complexity and real-time processing capabilities. The findings guide the selection of on-board computer models, thereby enhancing drone capabilities in critical operations © 2024 IEEE.

关键词： Drones

来源：评论

学校读者我要写书评

暂无评论

Exploring the Potential of MANETs: A Study on Raspberry Pi and UAV Integration 3

Exploring the Potential of MANETs: A Study on Raspberry Pi a...

引用

3rd International Conference on Intelligent Data Communication Technologies and Internet of Things, IDCIoT 2025

作者： Archana, B. Mahalakshmi, A. Revathi, K. Ashok Kumar, K. Hema, A.B. Dept. of Computer Science and Engineering Sri Ramakrishna Engineering College Coimbatore India Dept. of Information Technology SNS College of Engineering Coimbatore India

ISBN: (数字)9798331527549

ISBN: (纸本)9798331527549

This research investigates the feasibility of utilizing Mobile Ad Hoc Networks (MANETs) in conjunction with Raspberry Pi-equipped Unmanned Aerial Vehicles (UAV) swarms. The primary objective is to overcome the limitations of traditional communication infrastructures in dynamic and decentralized environments. Specifically, we address the challenge of establishing reliable and efficient communication links within a swarm of UAVs engaged in various applications, including disaster response, military reconnaissance, and environmental monitoring. By integrating Raspberry Pi devices with UAVs, we aim to create a self-organizing communication network that enables seamless data transmission, coordinated swarm operations, and enhanced mission effectiveness. The research explores different synchronization strategies, including master-slave and decentralized approaches, to optimize swarm coordination. Furthermore, we investigate the potential of incorporating advanced technologies such as machine learning, computer vision, and swarm intelligence to further enhance the autonomy, adaptability, and overall performance of the UAV swarm. © 2025 IEEE.

关键词： Mobile ad hoc networks

来源：评论

学校读者我要写书评

暂无评论

Land Cover Extraction from Satellite Imagery using U-Net Model 18

Land Cover Extraction from Satellite Imagery using U-Net Mod...

引用

18th INDIAcom;11th International Conference on Computing for Sustainable Global Development, INDIACom 2024

作者： Rani, K. Pushpa Sreedevi, Pogula Dastagiriah, C. Rao, K. Srinivas Subbaraju, G.V. Padmavathi, Jasti Mlr Institute of Technology Dept. of Computer Science and Engineering Hyderabad India G. Pulla Reddy Engineering College Dept. of Computer Science and Engineering Kurnool India Anurag University Dept. of Computer Science and Engineering Hyderabad India Scetw Dept. of Computer Science and Engineering Hyderabad India Hyderabad India

ISBN: (纸本)9789380544519

Satellite imagery offers extensive information that can be used for a variety of societal applications, from the number of buildings in a metropolis to the land cover types of a specific area. However, extracting such information is a difficult task. In the past, images were scanned manually by domain experts to extract features like buildings, infrastructure, and vegetation, which would take weeks or even months depending on the image size. To address this challenge, we trained our model using U-Net architecture, a deep learning model, because of its advantages in multiscale feature learning and faster processing. Data augmentation is done on the dataset in order to enhance the amount of data and prevent overfitting artificially. The trained model achieved an F1 score of 92% and an impressive precision score of 95%. This automated method can be scaled to large sets of data, reducing the need for human labeling. © 2024 Bharati Vidyapeeth, New Delhi.

关键词： Data augmentation Deep learning Machine learning Multiscale feature learning U-Net architecture

来源：评论

学校读者我要写书评

暂无评论

NLP Powered Restaurant Chatbot for Exceptional Customer Engagement 2

NLP Powered Restaurant Chatbot for Exceptional Customer Enga...

引用

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

作者： Goyal, Diya Guru Prasad, M.S. Garg, Ishika Tewari, Shashank Pandu N, K. Asha, P.N. Dept. of Computer Science and Engineering Dehradun India Srinivas University Institute of Engineering & Technology Dept. of Computer Science and Engineering Surathkal India REVA University School of Computer Science and Engineering Bengaluru India

ISBN: (纸本)9798350371314

technology is very volatile in nature, in a way that it keeps on changing daily. Restaurants are no different in the effects of this changing scenario. The primary objective of this research paper is to explore the historical trajectory of restaurants, analyze the current trends shaping the industry, and delve into the potential of chatbot integration to overcome existing challenges. Chatbots can play a pivotal role in enhancing the performance of businesses as per the research done. An integrated Dialog Flow application has been proposed, incorporating Rule-based grammar matching and M.L matching algorithms. This proposed work is trying to make NLP-based chatbots which may expedite the endeavor of businesses to produce and become profitable. © 2024 IEEE.

关键词： Python

来源：评论

学校读者我要写书评

暂无评论

Real-Time Object Detection from Surveillance using Deep Learning 3

Real-Time Object Detection from Surveillance using Deep Lear...

引用

3rd International Conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics, IITCEE 2025

作者： Tanisha, A. Tanisha, N. Chaitra, M. Tejasri, P.R. Bnm Institute of Technology Vtu Dept. of Computer Science and Engineering Bangalore India

ISBN: (数字)9798331515911

ISBN: (纸本)9798331515911

Object detection in surveillance systems leverages advanced deep learning techniques to enhance security measures through real-time analysis of dynamic video feeds. This project integrates the YOLOv5 model for detecting weapons in both pre-recorded videos and live camera feeds. The model, trained on a dataset with 4000 images labeled for handguns and knives, utilizes image preprocessing steps such as resizing, normalization, and augmentation to improve detection accuracy. Implementing the system with OpenCV and Tkinter, the application processes video streams, identifying potential threats with high precision rate of 0.85 and recall rate of 0.90. The integration of a graphical user interface ensures user-friendly operation for security personnel. Key outcomes demonstrate the efficacy of deep learning models in real-time object detection, significantly contributing to proactive surveillance and enhanced situational awareness. © 2025 IEEE.

关键词： Graphical user interfaces

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：