检索结果-内蒙古大学图书馆

Lightweight image super-resolution reconstruction based on mixed attention and global inductive bias network

Multimedia Tools and Applications 2024年 1-17页

作者： Cai, Yuxi Wen, Xiaopeng Wang, Tongguan College of Computer Science Sichuan University Chengdu China Department of Electronic Information and Electrical Engineering Dalian University of Technology Dalian China College of Informatics Huazhong Agricultural University Wuhan China

Convolutional neural network has significantly advanced the field of image super-resolution reconstruction in recent years. The insufficient ability to model global information of hierarchical features, incomplete attention to information, and excessive parameters make some based on convolutional methods face challenges. To address these issues, this paper designs the lightweight image super-resolution reconstruction based on Mixed Attention and Global Inductive bias Network(MAGIN), which directly captures the global information of phased features from the horizontal and vertical directions respectively to produce global inductive bias information. Furthermore, the global spatial attention block is proposed, which can realize the perception of global feature space without redundant operation. To realize more comprehensive attention for features, a mixed attention block arises, which enables the network to attend to features across different dimensions: channel, spatial, and pixel levels. Extensive experimental results demonstrate that the MAGIN delivers more satisfactory results than some other advanced lightweight reconstruction algorithms on several standard test datasets under the condition of considering the network’s performance, parameter and computational complexity. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

ICEL: Learning with Inconsistent Explanations 48

ICEL: Learning with Inconsistent Explanations

引用

48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023

作者： Liu, Biao Wu, Xiaoyu Yuan, Bo Southern University of Science and Technology Department of Computer Science & Engineering China Rams Lab Huawei China

ISBN: (纸本)9781728163277

Generating the heatmaps is one of the explanation methods to show what regions the model use to predict in vision tasks. GradCAM is a popular approach to provide such heatmaps. However, GradCAM is post-hoc, and its heatmaps are not always meeting human annotations. Inspired by CGC (Contrastive GradCAM consistency), we propose ICEL (InConsistent Explanation Learning) method which introduces inconsistent explanation loss measured by cosine similarity on heatmaps. We show that our method can preserve classification accuracy, while the heatmaps generated by explanation methods are more consistent with human annotations, and the computational complexity is reduced from O(n2) to O(n), compared with CGC. © 2023 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Effectiveness of Almost Sub Frame Approach as Inter Cell Interference Management in LTE-A Macro - Pico Scenario 9

Effectiveness of Almost Sub Frame Approach as Inter Cell Int...

引用

9th IEEE International Conference for Convergence in technology, I2CT 2024

作者： Jember, Kehali A. Abose, Tadele A. Kassahun, Ermias Madhav, T. Venu Adama Science And Technology University Department Electronics And Communication Engineering Adama Ethiopia Mettu University Department Electrical And Computer Engineering Ethiopia Gondar University Department Electrical And Computer Engineering Gondar Ethiopia

ISBN: (纸本)9798350394474

Heterogeneous networks are promising solutions for enhancing network performance of LTE-A mobile networks by deploying small cells within the area of the serving macro cells. The goal of deploying such networks is to offload data from the congested macro cells towards the small cells. However, cross tier interference is a serious issue, which is introduced due to the transmitted power difference between Macro cell node and small cell node. Hence, cell age users are mainly affected by high power interference originated from Macro node. The paper develops enhanced inter-cell interference coordination using Almost Blank Sub-Frame (ABS) to improve user throughput. This can be achieved by stopping data transmission on in certain sub frame of macro cell while Pico users are scheduled at the same frame. computer simulation results depict that the performance of Pico cell users is highly degraded, especially for cell edge users. The ABS technique improves cell edge user throughput of Pico cell by 95% and center cell user throughput by 75.25%. © 2024 IEEE.

关键词： Heterogeneous networks

来源：评论

学校读者我要写书评

暂无评论

Using Real-Time Integrated computer Vision and Deep Learning for Advanced Factory Safety 4

Using Real-Time Integrated Computer Vision and Deep Learning...

引用

4th IEEE International Conference on Electronic Communications, Internet of Things and Big Data, ICEIB 2024

作者： Tsai, Shang-Te Chang, Yu-Cheng Huang, Chao-Yi Yang, Fang-Yuan Chien, Wei Wu, Zong-Rong School Art and Design Sanming University Fujian China National Taipei University of Technology Department of Computer Science and Information Engineering Taipei Taiwan National Yunlin University of Science and Technology Department of Information Management Yunlin Taiwan Lunghwa University of Science and Technology Department of Computer Information and Network Engineering Taoyuan Taiwan National Yunlin University of Science and Technology Department of Technical and Vocational Program in Artificial Intelligence Yunlin Taiwan

ISBN: (纸本)9798350360721

Innovative safety in the workplace is vital as the high safety risks associated with electrical engineering construction can lead to injuries or even fatalities. Using computer vision technology, we experimented with scenarios such as 'normal operations' and 'unexpected incidents' to enhance safety measures. We integrated an Internet of Things system into the setup, enabling the system to quickly detect and alert to unexpected events in real-time, thereby improving workplace safety. © 2024 IEEE.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

Uni4DAL: A Unified Baseline for Multi-dataset 4D Auto-Labeling 27th

Uni4DAL: A Unified Baseline for Multi-dataset 4D Auto-Label...

引用

27th International Conference on Pattern Recognition, ICPR 2024

作者： Yang, Zhiyuan Wang, Xuekuan Zhang, Wei Tan, Xiao Lu, Jinchen Wang, Jingdong Ding, Errui Lai, Zhihui Zhao, Cairong Department of Computer Science and Technology Tongji University Shanghai201804 China Baidu Inc. Beijing China College of Computer Science and Software Engineering Shenzhen University Shenzhen518060 China

ISBN: (纸本)9783031781124

The 4D auto-labeling system, with its potential to enhance data annotation efficiency for 3D object detection, has garnered significant attention. However, its adoption has been hampered by the high costs associated with temporally annotated long-sequential training data and limited generalization capabilities across diverse scenarios. In this paper, we hypothesize that a multi-dataset approach can address these challenges and, accordingly, first introduce a Unified training pipeline for multi-dataset 4D Auto-Labeling, namely Uni4DAL. We recognize that this is a challenging task, primarily due to data-level variations and feature-level inconsistencies among various datasets. Motivated by this understanding, we initially propose a series of Data-Level Alignment (DLA) operations to mitigate potential discrepancies between diverse datasets and ensure synchronized training progress across samples from multiple datasets. Furthermore, to address feature-level inconsistencies, we introduce the Mixed Expert Models Voxel Feature Encoding (MoE-VFE) module, which aims to extract both domain-specific and domain-generalizable features. Additionally, we employ a Domain-Adaptive Hard Example Mining (DA-HEM) technique to leverage both data-level and feature-level consistencies, ensuring that the model pays enhanced attention to the challenging samples during multi-dataset training. Finally, comprehensive experiments demonstrate that Uni4DAL significantly improves performance on the nuScenes, Argoverse2, and Waymo datasets, and exhibits greater robustness with insufficient training data. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： 4D auto-labeling Multi-dataset training Offline 3D object detection

来源：评论

学校读者我要写书评

暂无评论

DHive: Query Execution Performance Analysis via Dataflow in Apache Hive 49th

DHive: Query Execution Performance Analysis via Dataflow in ...

引用

49th International Conference on Very Large Data Bases, VLDB 2023

作者： Zhang, Chaozu Shen, Qiaomu Tang, Bo Department of Computer Science and Engineering Southern University of Science and Technology China Research Institute of Trustworthy Autonomous Systems Southern University of Science and Technology China

Nowadays, Apache Hive has been widely used for large-scale data analysis applications in many organizations. Various visual analytical tools are developed to help Hive users quickly analyze the query execution process and identify the performance bottleneck of executed queries. However, existing tools mostly focus on showing the time usage of query sub-components (jobs and operators) but fail to provide enough evidence to analyze the root reasons for the slow execution progress. To tackle this problem, we develop a visual analytical system DHive to visualize and analyze the query execution progress via dataflow analysis. DHive shows the dataflow during query execution at multiple levels: query level, job level and task level, which enable users to identify the key jobs/tasks and explain their time usage by linking them to the auxiliary information such as the system configuration and hardware status. We demonstrate the effectiveness of DHive by two cases in a production cluster. © 2023, VLDB Endowment. All rights reserved.

关键词： Search engines

来源：评论

学校读者我要写书评

暂无评论

引用

23rd International Conference on Control, Automation and Systems, ICCAS 2023

作者： Amirullaeva, Sevara Han, Ji-Hyeong Seoul National University of Science and Technology Department of Computer Science and Engineering Seoul01811 Korea Republic of

ISBN: (纸本)9788993215274

The facial age estimation is an important task for efficient human-robot interaction. A supervised similarity learning has not been widely adopted to a facial age estimation and is still a challenging task. In this paper, we propose a multi-scale similarity learning model which is able to jointly learn similarities between multi-level age features extracted from three different images: an input, a sample with same age named positive and a sample with contrary age named negative. The idea behind using the multi-level feature extraction is to make our feature similarity learning model robust to scale invariance of individual images. As is known, the quality of images in datasets for age estimation vary considerably and deep-learning based networks are often sensitive to image quality and resolution. On that account, we employ a multi-scale feature extraction structure to our model and prove its ability to extract reliable features for a similarity learning. Experimental results show that the proposed approach outperforms previous research and provides a new state-of-the-art age estimation accuracy of UTKFace and CACD benchmark datasets for age estimation. © 2023 ICROS.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Enhanced Classification of Intel Images Using Refined EfficientNet and MobileNetV2 Frameworks 4

Enhanced Classification of Intel Images Using Refined Effici...

引用

4th International Conference on Intelligent Technologies, CONIT 2024

作者： Jain, Raghav Sharma, Rishabh Tiwari, Devesh Joshi, Kireet Jain, Vishal Chitkara University Institute of Engineering and Technology Chitkara University Punjab India Graphic Era Hill University Computer Science and Engineering Uttarakhand Dehradun248002 India Graphic Era Deemed to Be University Computer Science & Engineering Uttrakhand Dehradun248002 India Sharda University Sharda School of Engineering and Technology Department of Computer Science and Engineering U.P Greater Noida India

ISBN: (纸本)9798350349900

The proposed study describes the functioning of two modern deep learning models, namely EfficientNet and MobileNetV2, within the scope of image classification from a wide range of datasets. This study is designed to categorize images into six groups, namely: buildings, mountains, forests, sea/ocean, glaciers, and streets. The study utilizes a dataset from the Kaggle repository to assess the effectiveness of these models. The research methodology will include data preprocessing, modeling, and model evaluation. This paper will find out the balance between accuracy and efficiency which is provided by each model. EfficientNet is compared with MobileNetV2 because of its scalable structure and compromise between model depth, width, and resolution versus the latter which is known for its low weight and efficiency, especially in constrained circumstances. The research utilizes a framework of accuracy, precision, recall, F1-score metrics, and model efficiency analyses for comprehensive result evaluation and scrutiny. The analysis of this study indicates the unique advantages of separate models, for instance, EfficientNet is more accurate in most cases and MobileNetV2 is a good solution when you have to save computer resources. Comparing models highlights the importance of the selection that is made based on the specific application instead of performance metrics or computational demands. © 2024 IEEE.

关键词： Deep Learning EfficientNet Intel MobileNetV2

来源：评论

学校读者我要写书评

暂无评论

Multi lingual Video Summarizer using Natural Language Processing 5

Multi lingual Video Summarizer using Natural Language Proces...

引用

5th IEEE International Conference for Emerging technology, INCET 2024

作者： Adithya, T. Sai Divya, P. Vamshiteja, N. Prasad, Andraju Bhanu Hariharan, Shanmugasundaram Kukreja, Vinay Department of Computer Science and Engineering Vardhaman College of Engineering Hyderabad India Chitkara University Institute of Engineering and Technology Punjab India

ISBN: (纸本)9798350361155

Throughout the day, a huge quantity of video recordings is made and shared online. Because most films are uploaded with little to no context, it has become quite difficult to spend time watching these kinds of movies in today's hectic lifestyle. Our efforts may also be in vain if we are unable to find any useful information from them. The quantity of videos that need to be made could be directly impacted by the sharp rise in YouTube users' year over year. The likelihood that the makers of the videos will provide inaccurate information about the original content of the film is certainly very high due to their desperation for views. The user's precious time and resources could be wasted in this way. A summary button is one of the user-friendly features of the summarizer Chrome extension, which is designed to further enhance user interaction. This integration of Gensim algorithm with the summarizer Chrome extension streamlines the browsing experience, helping users navigate the overwhelming amount of online video content effectively. When you click this button, the Google Chrome web browser's Chrome extension shows the synopsis of the YouTube video that is currently playing. © 2024 IEEE.

关键词： Video recording

来源：评论

学校读者我要写书评

暂无评论

Dynamic Container Optimization for Efficient Microservices Deployment

Dynamic Container Optimization for Efficient Microservices D...

引用

2024 International Conference on Electrical, Electronics and Computing Technologies, ICEECT 2024

作者： Tiwari, Utkarsh Rai, Srishti Gaur, Ishu Khurana, Shiraz Sharda University Sharda School of Engineering and Technology Department of Computer Science and Engineering Uttar Pradesh Greater Noida India

ISBN: (纸本)9798350378092

This study presents a novel technique to Dynamic Container Optimization that aims to improve the deployment efficiency of microservices in cloud settings. The strategy uses a heuristic technique for intelligent microservice-container pairing and is focused on reducing operational expenses, guaranteeing Quality of Service (QoS) compliance, and resolving the container cold start issue. By methodically choosing the best matches between microservices and containers, the algorithm reduces resource waste and maximizes CPU and memory usage. To address the issue of cold starts in containers, the study also suggests a reusability method for containers that is predicated on predetermined principles and strategically identifies warm containers. The algorithm's thorough performance analysis demonstrates consistent resource utilization, quick processing times, and substantial cost-effectiveness, highlighting the Dynamic Container Optimization technique as a viable answer to the changing microservices deployment challenges in dynamic cloud environments. © 2024 IEEE.

关键词： Quality of service

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：