检索结果-内蒙古大学图书馆

5th International Conference on deep learning, Artificial Intelligence and Robotics, ICDLAIR 2023

作者： Tiwari, Kanishka Bhatia, Anmolpreet Singh Garg, Nibhrit Arora, Ishaan Saini, Poonam Department of Computer Science and Engineering Punjab Engineering College Chandigarh India

ISBN: (纸本)9783031609343

The utilization of web-based applications has greatly increased in recent years. This, in turn, has raised security concerns and data breaches leading to losses of millions of dollars. A large part of these attacks is carried out through modifications in HTTP requests and are often too complex to be recognised through simple rule-based checks. Thus, it is vital to utilize machine learning approaches in order to protect the applications against such attack vectors. Additionally, given the pace at which web servers need to operate, these machine learning methods need to be extremely fast at processing the requests. The paper presents a comparison of two recent deep learning models to recognize the web application attacks namely, a convolution neural network (CNN) and a Transformer-based model. We compare the prediction accuracies of both the models as well as their prediction time to identify the best model suited for being deployed in a real-life scenario. Based on implementation and results, it is observed that for our dataset, CNN tends to perform better in terms of both, accuracy and prediction time. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

关键词： Adversarial machine learning

来源：评论

学校读者我要写书评

暂无评论

Accurate and Efficient Loop Closure Detection With deep Binary image Descriptor and Augmented Point Cloud Registration

Accurate and Efficient Loop Closure Detection With Deep Bina...

引用

2024 International Conference on Intelligent Robots and Systems

作者： Wang, Jialiang Gao, Zhi Lin, Zhipeng Zhou, Zhiyu Wang, Xiaonan Cheng, Jianhua Zhang, Hao Liu, Xinyi Chen, Ben M. Wuhan Univ Sch Remote Sensing & Informat Engn Wuhan 430072 Peoples R China Chinese Univ Hong Kong Dept Mech & Automat Engn Hong Kong 999077 Peoples R China Wuhan Univ Hubei Luojia Lab Wuhan 430079 Peoples R China ZG Technol Co Ltd Wuhan 430000 Peoples R China

ISBN: (纸本)9798350377712;9798350377705

Loop Closure Detection (LCD) is an essential component of Simultaneous Localization and Mapping (SLAM), helping to correct drift errors, facilitate map merging, or both by identifying previously observed scenes. Despite its importance, traditional LCD algorithms based on single sensor such as camera or LiDAR exhibit degraded performance in challenging scenarios due to their inherent limitations. To address this issue, we propose a novel LCD method based on camera-LiDAR fusion, exploiting the rich textural information from cameras and the accurate geometric data from LiDAR to ensure robustness and speed in challenging environments. Specifically, we first employ deep hashing learning to encode deep image features into binary image descriptors for extremely fast loop candidate (LC) retrieval. Then, LiDAR points are augmented with image color for accurate geometric verification. Finally, we incorporate a spatial-temporal consistency check that mandates an LC to have consistently matched neighbors to be accepted as true. Our method is extensively verified and compared with the state-of-the-art methods on various datasets encompassing both indoor and outdoor environments. Experimental results demonstrate that our method obtains the best performance, increasing the maximum recall rate at 100% precision by a significant margin of 20% while operating in real-time at an average speed of 30 fps.

关键词： Binary images

来源：评论

学校读者我要写书评

暂无评论

Personalized Recommendation of Teacher learning Behavior Using Expression Recognition Algorithms Integrated by deep learning and Neural Networks 4

Personalized Recommendation of Teacher Learning Behavior Usi...

引用

4th International Signal processing, Communications and Engineering Management Conference, ISPCEM 2024

作者： Si, Chen Fangna, Chen Wanbing, Ma Xiaolei, Wang Jiahua, Zhu Nini, Cheng School of Art and Technology Chengdu College of UESTC Sichuan Chengdu613000 China College of Education Linyi University Shandong Linyi276000 China School of Life Sciences Central China Normal University Hubei Wuhan430079 China Zhengzhou Fengyang Foreign Language School Henan Zhengzhou450000 China Center for International Education Philippine Christian University Manila1004 Philippines School of Life Sciences Linyi University Shandong Linyi276000 China

ISBN: (纸本)9798331528676

At present, online teacher training platforms represented by Chinese university catechism have become an important way for teachers' professional development. However, the course dropout rate of such platforms is very high. For this reason, this paper intends to utilize a combination of deep learning and augmented learning to monitor the teacher's facial expression in the process of online teaching in real time, so as to realize the analysis of the teacher's learning behavior and emotional state. This paper proposes to investigate a facial expression recognition method based on a combination of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), using augmented learning strategies to train and optimize the proposed model to better understand the teacher's emotional response and learning needs. In this paper, we use machine learning and other methods to study teachers' emotional changes in the specific teaching process and conduct correlation analysis with them to find out the important factors affecting their learning effectiveness. When the teacher's emotion reaches a pleasant state, the learning completion rate is significantly increased by 20%, indicating that the emotional state has a significant positive impact on learning behavior and performance. Through the in-depth analysis of teachers' facial expressions and learning behaviors, we can optimize the appropriateness and engagement of the training content, as well as provide new research ideas and methods for educational technology research and promote the development of educational informatization. ©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

Expoliting Confidence-Based Model Fusion for Boosting image Classification Accuracy 2

Expoliting Confidence-Based Model Fusion for Boosting Image ...

引用

2nd International Conference on image processing, Computer Vision and Machine learning, ICICML 2023

作者： Jiang, Xinjian Nanjing University Department of Computer Science and Technology Jiangsu Nanjing210008 China

ISBN: (纸本)9798350331417

In the realm of deep learning, the traditional approach has been to train specialized models for individual tasks, which, although effective, is resource-intensive. The advent of large, universal models has mitigated this issue by offering multitask capabilities, reduced training time, and lower computational costs. However, these generalized models often underperform on specific tasks compared to specialized models. This paper introduces an innovative ensemble approach that integrates specialized and generalized models, specifically focusing on Contrastive Language-image Pretraining (CLIP) and EfficientNet. This work proposes three fusion strategies: Weighted Voting, Confidence Comparison, and Fully Connected Network Fusion, and evaluate them using the CIFAR-100 dataset. The ensemble model significantly outperforms individual models, achieving an adjusted accuracy of up to 0.848. The paper also introduces a novel evaluation metric, Confidence-Accuracy Correlation, to assess the reliability of model confidence. The findings could revolutionize ensemble learning by making it more adaptive and suited for real-world applications, thereby pushing the boundaries of possibility in artificial intelligence. © 2023 IEEE.

关键词： CLIP deep learning image classification

来源：评论

学校读者我要写书评

暂无评论

Analysis of Onground to Onboard processing Capabilities for Flood Segmentation using deepLabv3+ 3

Analysis of Onground to Onboard Processing Capabilities for ...

引用

3rd International Conference for Innovation in Technology, INOCON 2024

作者： Varshit, Namburu Chaitanya Ramesh Kumar, P. Rishi, Gudimetla Hanumath V.R.Siddhartha Engineering College Department of CSE Vijayawada India

ISBN: (纸本)9798350381931

Flood segmentation is a crucial aspect of disaster management and remote sensing applications. deep learning models have shown remarkable proficiency in semantic segmentation tasks. However, adapting these models for flood segmentation within an onboard processing framework presents unique challenges and opportunities. This analysis focuses on onground and onboard capabilities, carefully examining the advantages and implications of using single-board computers for real-time processing. The shift from onground to onboard processing, though promising, requires careful consideration of resource constraints. Onboard deployment can greatly improve the speed of flood detection and response, enabling faster and more informed decision-making. Additionally, it reduces dependence on external computational resources, making the system more resilient in disaster-prone areas where infrastructure may be compromised. To compare the performance of flood segmentation onground and onboard, the processing time, memory usage, and CPU consumption can be measured. Based on the results, the analysis led to the conclusion that the model produced an accuracy of 82% on a single-board computer. © 2024 IEEE.

关键词： Floods

来源：评论

学校读者我要写书评

暂无评论

Autonomous Robot Navigation: deep learning Approaches for Line Following and Obstacle Avoidance 20

Autonomous Robot Navigation: Deep Learning Approaches for Li...

引用

20th CSI International Symposium on Artificial Intelligence and Signal processing, AISP 2024

作者： Alitappeh, Reza Javanmard Mahmoudi, Nima Jafari, Mohammad Reza Foladi, Ali University of Science and Technology of Mazandaran Department of Electrical and Computer Engineering Behshahr Iran

ISBN: (纸本)9798350383942

This research presents a navigation robotic system designed for the concurrent tasks of line following and obstacle avoidance in partially-known environments with presence of obstacles. By applying a strategically positioned camera for precise line following with a LSTM model and distance sensors guided by a CNN model for obstacle avoidance, our system exhibits robust performance. The seamless transition between these modes, driven by real-time environmental inputs, underscores the adaptability and autonomousness of the platform. Experimental results indicate a notable enhancement in performance, demonstrating the efficacy of the proposed approach in achieving superior outcomes in challenging robotic navigation scenarios. © 2024 IEEE.

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

ECCNet: Efficient chained centre network for real-time multi-category vehicle tracking and vehicle speed estimation

引用

IET INTELLIGENT TRANSPORT SYSTEMS 2022年第11期16卷 1489-1503页

作者： Yu, Chaoshun Yang, Jianxi Jiang, Shixin Zhang, Yufan Li, Hao Du, Lifang Chongqing Jiaotong Univ Coll Traff & Transportat 66 Xuefu Ave Chongqing Peoples R China Chongqing Jiaotong Univ Sch Informat Sci & Engn 66 Xuefu Ave Chongqing 400074 Peoples R China

To achieve the intelligent perception of traffic video and overcome the dispersion of computing modules and the separation processing of multiple tasks, efficient chained centre network (ECCNet) is proposed as a unified framework, accomplishing detection, vehicle classification, tracking, and vehicle speed estimation simultaneously. First, for the speed-accuracy trade-off, CA-CenterNet is presented, which can detect vehicles and classify vehicle types to serve cross-frame tasks more accurately by embedding coordinate attention. Second, 3D convolution is employed to construct a self-adaptive branch for data association and speed estimation, respectively. This self-adaptive approach leverages the power of deep learning to enhance tracking performance and avoid camera calibration via capturing motion information across frames. Moreover, a chained structure is adopted to reuse the backbone feature map. The spatio-temporal information of adjacent frames can be extracted at almost no additional cost. Finally, the above single-frame and cross-frame tasks are integrated into a unified multi-task collaborative optimization model. The effectiveness of ECCNet is verified with experiments on the UA-DETRAC dataset. ECCNet achieves 55.5% MOTA, 0.76 F1 value, and 3.10 MAE with an inference speed of 32.5 Hz in the tracking, classification, and speed estimation tasks, respectively.

关键词： cross-frame tasks speed estimation tasks UA-DETRAC dataset vehicle speed estimation object tracking chained structure road vehicles camera calibration deep learning (artificial intelligence) image classification optimisation cameras single-frame task speed-accuracy trade-off convolutional neural nets road traffic efficient chained centre network self-adaptive approach vehicle detection CA-CenterNet object detection deep learning spatio-temporal information data association traffic video feature extraction self-adaptive branch 3D convolution vehicle type classification adjacent frame extraction vehicle tracking tracking performance enhancement motion information ECCNet real-time multicategory vehicle tracking image fusion image motion analysis coordinate attention embedding inference speed video signal processing backbone feature map intelligent perception traffic engineering computing unified multitask collaborative optimization model estimation theory

来源：评论

学校读者我要写书评

暂无评论

MultiPanoWise: holistic deep architecture for multi-task dense prediction from a single panoramic image

MultiPanoWise: holistic deep architecture for multi-task den...

引用

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Shah, Uzair Tukur, Muhammad Alzubaidi, Mahmood Pintore, Giovanni Gobbetti, Enrico Househ, Mowafa Schneider, Jens Agus, Marco Hamad Bin Khalifa Univ ICT Div Coll Sci & Engn Doha Qatar CRS4 Visual & Data Intens Comp Pula Italy Natl Res Ctr HPC Big Data & Quantum Comp Pula Italy

ISBN: (纸本)9798350365474

We present a novel holistic deep-learning approach for multi-task learning from a single indoor panoramic image. Our framework, named MultiPanoWise, extends vision transformers to jointly infer multiple pixel-wise signals, such as depth, normals, and semantic segmentation, as well as signals from intrinsic decomposition, such as reflectance and shading. Our solution leverages a specific architecture combining a transformer-based encoder-decoder with multiple heads, by introducing, in particular, a novel context adjustment approach, to enforce knowledge distillation between the various signals. Moreover, at training time we introduce a hybrid loss scalarization method based on an augmented Chebychev/hypervolume scheme. We illustrate the capabilities of the proposed architecture on publicdomain synthetic and real-world datasets. We demonstrate performance improvements with respect to the most recent methods specifically designed for single tasks, like, for example, individual depth estimation or semantic segmentation. To our knowledge, this is the first architecture capable of achieving state-of-the-art performance on the joint extraction of heterogeneous signals from single indoor omnidirectional images.

关键词： dense estimation indoor environments multi-task learning panoramic images

来源：评论

学校读者我要写书评

暂无评论

A fiber optic sensing intrusion detection method based on WPD-EMD and improved ResNet

引用

OPTICAL FIBER TECHNOLOGY 2025年 90卷

作者： Hu, Xinmin Bai, Xinrui Li, Jingqi He, Yiting Li, Yingying Li, Liang Xiao, Han Liu, Cong Zhang, Fan Tang, Jing Hu, Sheng Hubei Univ Technol Sch Elect & Elect Engn Wuhan 430068 Peoples R China Hubei Univ Technol Hubei Key Lab High Efficiency Utilizat Solar Energ Wuhan 430068 Peoples R China Hubei Univ Technol Xiangyang Ind Inst Xiangyang 441000 Peoples R China

Perimeter security systems are essential for safeguarding critical locations from unauthorized intrusions and various security threats. Traditional video surveillance has limitations like coverage gaps, blind spots, and the need for extensive manual analysis. In contrast, fiber optic systems use distributed sensing technology for realtime, precise data acquisition and automatic anomaly detection, reducing the need for manual monitoring. The Phase-Sensitive Optical time-Domain Reflectometer (Phi-OTDR) is noted for its high sensitivity, large dynamic range, and robust interference resistance, making it ideal for extensive real-time monitoring. To enhance recognition accuracy, especially for high-threat events requiring zero false alarms, this study proposes a fiber optic sensing signal recognition method using Wavelet Packet Decomposition (WPD) combined with Empirical Mode Decomposition (EMD) and an improved ResNet architecture. Encoding one-dimensional signals into images using Recurrence Plots (RP) leverages advanced techniques from image processing and computer vision, enhancing signal recognition accuracy and application scope. Experimental results show that the WPD-EMD denoising method significantly improves the quality of the original signals, achieving an overall recognition rate of 93.83% for eight types of intrusion signals. For the most representative events (background noise, excavator digging, truck passing, stone knocking), the recognition rate reaches 99.69%. This method shows significant potential for advancing perimeter security monitoring.

关键词： Phase-sensitive optical time domain reflectometry (Phi-OTDR) Signal denoising deep learning Event recognition Perimeter security

来源：评论

学校读者我要写书评

暂无评论

Hybrid path planning and following of a quadrotor UAV based on deep reinforcement learning 36

Hybrid path planning and following of a quadrotor UAV based ...

引用

36th Chinese Control and Decision Conference (CCDC)

作者： Zhong, Liangsheng Zhao, Jiasheng Luo, Haining Hou, Zhiwei Sun Yat Sen Univ Sch Syst Sci & Engn Guangzhou Peoples R China

ISBN: (纸本)9798350387780;9798350387797

deep reinforcement learning(DRL)-based path planning algorithms have gained significant attention in recent years due to their end-to-end processing and robustness. Nevertheless, they face challenges in scenarios with long distances and dense obstacles because they only consider local environmental information. This paper presents a hybrid path planning and following approach that combines path planning based on soft actorcritic(SAC) with path following. Initially, a path is generated using the sampling-based method Adaptively Informed Trees (AIT*), and the path subsequently serves as tracking points for the planner. This approach guides the agent to move faster and more effectively toward the goal, and continuously updates the tracking point in real time. We conducted an experiment to evaluate the training process and performance of this hybrid path planning approach based on DRL while comparing it to the original DRL-based approach. The experimental results illustrate the superiority of the presented approach in both the training process and task performance.

关键词： Path Planning Path Follow deep Reinforcement learning UAV Soft Actor-Critic

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：