Authentication email has already been sent, please check your email box: and activate it as soon as possible.
You can login to My Profile and manage your email alerts.
If you haven’t received the email, please:
|
|
There are 49 papers published in subject: since this site started. |
Select Subject |
Select/Unselect all | For Selected Papers |
Saved Papers
Please enter a name for this paper to be shown in your personalized Saved Papers list
|
1. KDText: A Lightweight Scene Text Detector with Decoupling-Based Knowledge Distillation | |||
Lei Siyue,Yan DanFeng | |||
Computer Science and Technology 28 February 2023 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:In recent years, scene text detection task has made great progress. However, most of works are devoted to improving the performance of detector and pay little attention to the practical applications. In this paper, we propose to exploit a mask branch to detect arbitrary-shaped text accurately, and compress the model to reduce computation and storage costs, achieving a balance between speed and accuracy. Specifically, we distill intermediate features and text proposal classification to transfer dark knowledge to student text detector. In the distillation process, we treat textual and Background features differently and decouple positive and negative text proposals. Experimental results on the ICDAR 2015 and ICDAR 2017 MLT datasets demonstrate the superiority of our lightweight scene text detector. | |||
TO cite this article:Lei Siyue,Yan DanFeng. KDText: A Lightweight Scene Text Detector with Decoupling-Based Knowledge Distillation[OL].[28 February 2023] http://en.paper.edu.cn/en_releasepaper/content/4759241 |
2. Multidimensional Features Based Model for Social Network User Classification | |||
MO Qin-Chu,DENG Xiao-Long1,SONG Lin-Ming | |||
Computer Science and Technology 12 March 2021 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:In recent years, social network users have been the focus of research, but existing research often divides users into normal and malicious users, lacking a more detailed analysis. To address the absence of a precise user classification model, this paper builds a three-dimensional classification model including Anti-bot Analysis, Publicity, and Hashtag Manipulation for measuring user behaviors and classifying users into four categories: harmless users, publicizing users, hashtag hijacking users, and malicious publicizing users. Based on the classification model, this paper has built a set of user classification features, and introduces four new features. We also create two new Weibo user datasets with new features, and re-processes an existing Weibo user dataset. The experimental results show that the model and features proposed in this paper have good classification efficiency for precise Weibo user classification and bot behavior recognition, and have obvious classification improvement for decision trees and BP neural networks. | |||
TO cite this article:MO Qin-Chu,DENG Xiao-Long1,SONG Lin-Ming. Multidimensional Features Based Model for Social Network User Classification[OL].[12 March 2021] http://en.paper.edu.cn/en_releasepaper/content/4753963 |
3. A Database Architecture of E-commerce System with High Availability and Scalability based on MySQL and Redis | |||
CUI Yansong,BAI Chunyu | |||
Computer Science and Technology 11 January 2021 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:With the continuous development of e-commerce systems, data storage has become a crucial technology in this era, and building a database environment with high-availability has become an urgent demand of many enterprises. In order to prevent web application crashes caused by database server downtime, this paper proposes a database architecture of e-commerce system with high availability and scalability based on MySQL and Redis. The architecture implements read-write separation and automatic switching when the server goes down to ensure high availability of e-commerce systems. It uses the sentinel to build Redis cluster, while reducing database access pressure and preventing cache invalidation caused by Redis server downtime. It builds a database cluster based on MySQL replication, which can ensure real-time synchronization and disaster recovery of data between database servers and improve scalability of system. Finally, this paper implemented an e-commerce system based on this architecture. | |||
TO cite this article:CUI Yansong,BAI Chunyu. A Database Architecture of E-commerce System with High Availability and Scalability based on MySQL and Redis[OL].[11 January 2021] http://en.paper.edu.cn/en_releasepaper/content/4753448 |
4. An improved Faster R-CNN network for aeroengine fuse fracture detection | |||
Liao Minjie,Bo Lin,Wu Xialing,Liu Qunyang,Wu Wenhong | |||
Computer Science and Technology 13 December 2020 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:In order to meet the needs of aeroengine fuse fracture detection in practical application, an improved Faster R-CNN small target detection network is proposed. Firstly, FPN feature graph pyramid is added to improve the extraction ability of small target features, and then ROI Align is used to replace ROI pooling to reduce the loss of feature information of small targets. Experiments on the fuse fracture data set show that the improved detection network is 5.76% higher than Faster R-CNN on mAP. The experimental results show that the improved network is more advanced and has a practical application prospect in aeroengine fuse fracture detection based on computer vision. | |||
TO cite this article:Liao Minjie,Bo Lin,Wu Xialing, et al. An improved Faster R-CNN network for aeroengine fuse fracture detection[OL].[13 December 2020] http://en.paper.edu.cn/en_releasepaper/content/4753217 |
5. Reducing Web Latency with Coding-Based Fast Multi-Path Loss Recovery | |||
LIU Yi,CHEN Guo | |||
Computer Science and Technology 14 May 2020 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:TCP latency is critical to the performance of web services. However, packet loss greatly impairs the TCP performance due to its poor loss recovery mechanisms. Recent work FUSO addressed this problem by leveraging multi-path diversity for proactive loss recovery, it used "good" paths to proactively retransmit the potentially lost packet on "bad" paths. However, because it's impossible to predict which packet is truly lost, FUSO tended to retransmit the oldest unACKed packet. Through analysis and comprehensive experiments, this paper shows that in the Internet scenario, such simple proactive retransmission of the oldest unACKed packet is not accurate enough to recover the lost packets, which causes performance penalty. To address the problem, this paper presents \name, a \fullname. Different from FUSO, when there is a chance for proactive loss recovery, \name generates a coding packet that codes all (or multiple) unACKed packets together. As such, \name can always proactively retransmit the ``right'' lost packet, since the receiver side can decode the lost packet by combining the coding packet with other received packets. \name is implemented in Linux kernel with \approx2K lines of code.Testbed and simulation results show that, under lossy condition, \name can greatly decrease the average and $99^{th}$ percentile flow completion time (FCT) by \approx12\% and \approx59\% in the testbed, and up to \approx16.9\% and \approx54.5\% in the simulation, respectively. | |||
TO cite this article:LIU Yi,CHEN Guo. Reducing Web Latency with Coding-Based Fast Multi-Path Loss Recovery[OL].[14 May 2020] http://en.paper.edu.cn/en_releasepaper/content/4752040 |
6. The Research on Web AR Avatar Generation System | |||
Feng Jingyi,Liao Jianxin | |||
Computer Science and Technology 13 February 2020 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:With the development of virtual reality and augmented reality (AR) technology, new types of social media continue to emerge and develop rapidly. One of the most important technologies of these social media applications is to generate a virtual image similar to that for users. Existing virtual image generation technologies are mainly divided into methods based on 3D scanning equipment and methods based on model library matching. Among them, the method based on the three-dimensional scanning device has high requirements on the device, which is not convenient to use and popularize; and the method based on the model library matching method has small individual differences in results. In order to solve the shortcomings of the existing methods, this paper proposes a method for directly generating a virtual image based on a two-dimensional image and generates an exclusive virtual image with personal characteristics for the user.This paper proposes a virtual image generation method based on web 3D face reconstruction and 3D model texture reconstruction. Use Tensorflow.js to transform the trained deep learning model into a format that can be recognized by the browser and predict the 3D face information and facial feature points on the web. Then, based on Delaunay triangulation and image affine transformation, a virtual image texture is generated for the model.The experimental results show that the method proposed in this paper can generate personalized and exclusive three-dimensional avatars based on the user's two-dimensional face information. It provides a feasibility reference for web-based virtual image related research. | |||
TO cite this article:Feng Jingyi,Liao Jianxin. The Research on Web AR Avatar Generation System[OL].[13 February 2020] http://en.paper.edu.cn/en_releasepaper/content/4750730 |
7. A Multi-feature Fusion Approach for Citation Recommendation | |||
Liu Ying | |||
Computer Science and Technology 21 November 2019 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:With the tremendous growth in the number of published academic papers, researchers find that it is a time-consuming task to search suitable references while writing scientific papers. To solve this problem, citation recommendation is proposed to recommend a list of candidate papers which are most relevant to the given manuscript. In this paper, we propose a context-aware citation recommendation approach based on BERT and side information. The proposed approach simultaneously incorporates sentence-level representation extracted by BERT from papers and network-based representation extracted by Node2Vec with joint-information (papers, authors and venues). Besides that, side information is taken into account when calculating the relevance between candidate papers and the target manuscript. By introducing the latest natural language processing algorithm and extracting rich features manually, the proposed approach performs well in precision and recall. When conducting experiment on AAN dataset, the results demonstrate the effectiveness of the proposed approach to optimize the quality of citation recommendation, compared with other baseline approaches. | |||
TO cite this article:Liu Ying. A Multi-feature Fusion Approach for Citation Recommendation[OL].[21 November 2019] http://en.paper.edu.cn/en_releasepaper/content/4750008 |
8. GranuleJ: A Context Check-based Programming Language for Flexible Runtime Adaptation | |||
ZENG Qing-Hua, ZHAO Yin-Liang, SUN Li-Yu, WU Wen-Bin | |||
Computer Science and Technology 18 December 2017 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:\renewcommand{\raggedright}{\leftskip=0pt \rightskip=0pt plus 0cm}\raggedrightModern applications tend to increasingly suffer from unpredicted context changes that may occur at any moment during the program execution, so it is urgently needed to adapt their behaviors to such frequently changing contexts dynamically. From the perspective of language-level, language extension is an efficient and prompt approach to conduct those adaptable applications. However, the existing context-based languages can only provide anticipated adaptation which is usually predefined at the initial design time, and they also lack appropriate programming language abstractions of dynamic flavor to support context uncertainty at runtime. In this paper, we present a novel programming language called \emph{GranuleJ}, which enables implicit context checks to be aware of the adaptation of the program and carry out program evolution when the program is no longer satisfied with the current context. GranuleJ introduces \emph{context variable} to identify context changes clearly, \emph{fitness tests} to detect the adaptation points where unsuitable program behaviors happen relying on context variables and \emph{granules} that modularize behavior variations as reuse building blocks to be freely assembled or disassembled at runtime. We have already implemented the language framework of GranuleJ and validated the feasibility and effectiveness of it through performance evaluation. | |||
TO cite this article:ZENG Qing-Hua, ZHAO Yin-Liang, SUN Li-Yu, et al. GranuleJ: A Context Check-based Programming Language for Flexible Runtime Adaptation[OL].[18 December 2017] http://en.paper.edu.cn/en_releasepaper/content/4742867 |
9. Facet Annotation by Extending CNN with a Matching Strategy | |||
Bei Wu, Bifan Wei, Jun Liu, Yuanhao Zheng, Zhaotong Guo, Qinghua Zheng | |||
Computer Science and Technology 08 May 2017 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:Most community question answering (CQA) websites manage plenty of question answer pairs (QAPs) through topic-based organization, which cannot satisfy users' search demands. Facets of topics serve as a powerful tool for navigating, refining, and grouping the QAPs.In this work, we propose FACM, a model for facet annotation by extending Convolution Neural Network (CNN) with a matching strategy. First, considering the importance of topic phrases for QAPs in knowledge domain, phrase information is incorporated into text representation by a CNN with different kernel sizes. Then, through a matching strategy among QAPs and fact label texts (FaLTs) acquired from external knowledge base, we generate similarity matrices to deal with facet heterogeneity. Finally, a three-channel CNN is trained for facet label assignment of QAPs as a binary classifier.Experiments on three real-world datasets show that FACM outperforms three state-of-the-art methods. | |||
TO cite this article:Bei Wu, Bifan Wei, Jun Liu, et al. Facet Annotation by Extending CNN with a Matching Strategy[OL].[ 8 May 2017] http://en.paper.edu.cn/en_releasepaper/content/4732079 |
10. Illumination and Rotation Invariant Featureof Texture Images Based on Hilbert-Huang Transform | |||
Yang Zhihua, Zhang Qian,Yang Lihua | |||
Computer Science and Technology 23 April 2017 | |||
Show/Hide Abstract | Cite this paper︱Full-text: PDF (0 B) | |||
Abstract:This paper presents a novel method to extract theillumination and rotation invariant features for texture imagesbased on Hilbert-Huang transform. Texture images are usually ofquasi-periodic. It is shown in this paper that the main frequencyof the Hilbert marginal spectrum of a texture image can be used to measure the approximate period effectively and thus can be servedas a good feature for texture classification. This feature isproved to be invariant to uneven illumination. Being modified, itis shown that this feature is also invariant rotation. Experimentshave been conducted to compare the feature with the existing ones.It is shown that the proposed approach outperforms the existingmethods in both recognition rate and robustness to unevenillumination, rotation and noise pollution. | |||
TO cite this article:Yang Zhihua, Zhang Qian,Yang Lihua. Illumination and Rotation Invariant Featureof Texture Images Based on Hilbert-Huang Transform[OL].[23 April 2017] http://en.paper.edu.cn/en_releasepaper/content/4726832 |
Select/Unselect all | For Selected Papers |
Saved Papers
Please enter a name for this paper to be shown in your personalized Saved Papers list
|
|
About Sciencepaper Online | Privacy Policy | Terms & Conditions | Contact Us
© 2003-2012 Sciencepaper Online. unless otherwise stated