Pub Date : 2020-07-01DOI: 10.1109/cisce50729.2020.00010
Xuanguang Chen
Urban computing is an emerging discipline to improve the quality of people’s life in city. This paper studies data processing which is one of the challenges of urban computing. The author compares different applications of urban computing using Graph Neural Network (GNN) for data processing, and draws the conclusion that GNN does have better results in the data processing of urban computing. By studying the different applications of GNN in urban computing, this paper shows the superiority of GNN in urban computing, and makes suggestions for the future application of GNN in urban computing at the same time.
{"title":"Application of GNN in Urban Computing","authors":"Xuanguang Chen","doi":"10.1109/cisce50729.2020.00010","DOIUrl":"https://doi.org/10.1109/cisce50729.2020.00010","url":null,"abstract":"Urban computing is an emerging discipline to improve the quality of people’s life in city. This paper studies data processing which is one of the challenges of urban computing. The author compares different applications of urban computing using Graph Neural Network (GNN) for data processing, and draws the conclusion that GNN does have better results in the data processing of urban computing. By studying the different applications of GNN in urban computing, this paper shows the superiority of GNN in urban computing, and makes suggestions for the future application of GNN in urban computing at the same time.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116915630","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The environment of the steel factory workshop is complex, and there may be a variety of unexpected potential dangers, so wearing a helmet to enter the workshop is a prerequisite for the factory. In order to supervise this situation, it is necessary for employees to wear helmets for testing, which is a key part of the overall intelligent monitoring system for steel plant personnel. In this paper, through the crawler to collect high-definition employees wearing helmets and no helmet pictures, using manual labeling, proposed a helmet detection framework based on computer vision deep learning detection framework Faster-RCNN. The actual testing results produce convincing experimental results, which proves the effectiveness and practicability of the proposed framework.
{"title":"Safety Helmet Wearing Detection Based on Image Processing and Deep Learning","authors":"Wei Zhang, Chifu Yang, Feng Jiang, Xianzhong Gao, Xiao Zhang","doi":"10.1109/CISCE50729.2020.00076","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00076","url":null,"abstract":"The environment of the steel factory workshop is complex, and there may be a variety of unexpected potential dangers, so wearing a helmet to enter the workshop is a prerequisite for the factory. In order to supervise this situation, it is necessary for employees to wear helmets for testing, which is a key part of the overall intelligent monitoring system for steel plant personnel. In this paper, through the crawler to collect high-definition employees wearing helmets and no helmet pictures, using manual labeling, proposed a helmet detection framework based on computer vision deep learning detection framework Faster-RCNN. The actual testing results produce convincing experimental results, which proves the effectiveness and practicability of the proposed framework.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124487840","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/CISCE50729.2020.00100
Jianqiu Chen
At present, the classification of vehicle brands as an important unit in the urban intelligent transportation system has become a hot spot for researchers from various countries. The classification and recognition of videos and images have been effectively researched and applied. In order to achieve better classification effect and higher recognition efficiency, this paper uses Principal Component Analysis-Net (PCA-NET) to realize the classification of vehicle brand, and combined with Support Vector Machines (SVM) to achieve. From the experimental results, this method can effectively extract the vehicle front view and achieve classification. The classification accuracy is high, which can reach 93.2%. In addition, this method has flexible adaptability to complex background conditions.
{"title":"Vehicle Brand Classification Method Based on PCA-NET Under Complex Background","authors":"Jianqiu Chen","doi":"10.1109/CISCE50729.2020.00100","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00100","url":null,"abstract":"At present, the classification of vehicle brands as an important unit in the urban intelligent transportation system has become a hot spot for researchers from various countries. The classification and recognition of videos and images have been effectively researched and applied. In order to achieve better classification effect and higher recognition efficiency, this paper uses Principal Component Analysis-Net (PCA-NET) to realize the classification of vehicle brand, and combined with Support Vector Machines (SVM) to achieve. From the experimental results, this method can effectively extract the vehicle front view and achieve classification. The classification accuracy is high, which can reach 93.2%. In addition, this method has flexible adaptability to complex background conditions.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133823332","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/cisce50729.2020.00087
K. Shi
With the continuous development of agriculture under the background of "Internet +", the original agricultural information service platforms need to be upgraded. According to the latest advances in intelligent agriculture, the design idea of network information service platform based on industry chain is put forward. The platform is positioned to the function of industry chain information link and multi-subject information exchange. A three-tier architecture is adopted in the platform, and the front-end is constructed by Angular, PWA and other technologies while the back-end is constructed by node. Js, Koa2 and other technologies, and many kinds of information service functions are designed in detail. This design idea provides help for the construction of other intelligent agricultural network information service platforms.
{"title":"Design of Network Information Service Platform for Intelligent Agricultural Industry Chain","authors":"K. Shi","doi":"10.1109/cisce50729.2020.00087","DOIUrl":"https://doi.org/10.1109/cisce50729.2020.00087","url":null,"abstract":"With the continuous development of agriculture under the background of \"Internet +\", the original agricultural information service platforms need to be upgraded. According to the latest advances in intelligent agriculture, the design idea of network information service platform based on industry chain is put forward. The platform is positioned to the function of industry chain information link and multi-subject information exchange. A three-tier architecture is adopted in the platform, and the front-end is constructed by Angular, PWA and other technologies while the back-end is constructed by node. Js, Koa2 and other technologies, and many kinds of information service functions are designed in detail. This design idea provides help for the construction of other intelligent agricultural network information service platforms.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134374639","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Cultural poverty and educational poverty are closely related. Without education, it is difficult to establish culture. Without culture, education will not be valued. Only by promoting cultural poverty alleviation and educational poverty alleviation as a whole, can we form a joint force to eliminate poverty culture and improve ideological, moral, scientific and cultural level. Only through high-level cultural poverty alleviation and educational poverty alleviation, can we really play the "supporting aspiration" of culture and the function of "supporting intelligence" to provide intellectual and human support for poverty-stricken areas to achieve industrial shaping and economic transformation, so as to achieve complete poverty alleviation. Therefore, Wish Magic Box will contact relevant educational institutions to conduct lectures and provide relevant support services. For example, in terms of service lectures, it will provide services that people need to work and solve employment problems to provide education support for people with education needs in poor areas.
{"title":"O2O Integrated AI Precise Poverty Alleviation Plan","authors":"Shih-Feng Chang, Hui Ding, Min-Qi Huang, Yong-Lin Tan, Jun-Jie Chen, Zhi-Tao Huang","doi":"10.1109/CISCE50729.2020.00096","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00096","url":null,"abstract":"Cultural poverty and educational poverty are closely related. Without education, it is difficult to establish culture. Without culture, education will not be valued. Only by promoting cultural poverty alleviation and educational poverty alleviation as a whole, can we form a joint force to eliminate poverty culture and improve ideological, moral, scientific and cultural level. Only through high-level cultural poverty alleviation and educational poverty alleviation, can we really play the \"supporting aspiration\" of culture and the function of \"supporting intelligence\" to provide intellectual and human support for poverty-stricken areas to achieve industrial shaping and economic transformation, so as to achieve complete poverty alleviation. Therefore, Wish Magic Box will contact relevant educational institutions to conduct lectures and provide relevant support services. For example, in terms of service lectures, it will provide services that people need to work and solve employment problems to provide education support for people with education needs in poor areas.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131319291","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/CISCE50729.2020.00092
Lipin Tan, Yu-tian Li
Automatic segmentation of plant images is an important step in the research on rape and weed recognition system. In order to overcome the shortcomings of traditional threshold method in rape gray image segmentation, an automatic segmentation method based on RGB color space was proposed. In RGB color space, the G component of vegetation is dominant, and the R component of soil background is dominant. A method combining G-0.8R color index with 0 threshold is proposed to segment rape image. Compared with traditional methods, this method has better segmentation effect and shorter processing time, which meets the real-time requirements of rape and weed identification system. Morphological operations and other denoising methods can further improve the image quality.
{"title":"Research on Rape Image Automatic Segmentation Method Based on RGB Color Space","authors":"Lipin Tan, Yu-tian Li","doi":"10.1109/CISCE50729.2020.00092","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00092","url":null,"abstract":"Automatic segmentation of plant images is an important step in the research on rape and weed recognition system. In order to overcome the shortcomings of traditional threshold method in rape gray image segmentation, an automatic segmentation method based on RGB color space was proposed. In RGB color space, the G component of vegetation is dominant, and the R component of soil background is dominant. A method combining G-0.8R color index with 0 threshold is proposed to segment rape image. Compared with traditional methods, this method has better segmentation effect and shorter processing time, which meets the real-time requirements of rape and weed identification system. Morphological operations and other denoising methods can further improve the image quality.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"77 10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131206181","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/CISCE50729.2020.00058
Xiaowei Ye, Ning Xu, Xiaofeng Liu, Xiao Yao, A. Jiang
The powerful capabilities of deep convolutional neural networks make them useful in various fields. However, most edge devices are difficult to afford the huge amount of parameters and high computational cost. Therefore, it is highly imperative to compress these huge models to make them lightweight to enable real-time inference on edge devices. Channel pruning is a mainstream method of network compression. Generally, the Lasso constraint is imposed on the scaling factor in the batch normalization layer to make them tend to zero for selecting unimportant channels and then prune them. However, Lasso is a non-smooth function that is not derivable at zero, we experimentally find that when the value of the loss function is small, it is difficult to decline continuously. Aiming at the above problems, this paper proposes a pruning strategy based on the derivable function Smooth-Lasso, using Smooth-Lasso as a regularization constraint to perform sparse training and then prune the network. Experiments on benchmark datasets and convolutional networks show that our method can not only make the loss function converge quickly, but also save more storage space and computational cost than the baseline method while maintaining the same level of accuracy as the original network.
{"title":"Efficient Network Compression Through Smooth-Lasso Constraint","authors":"Xiaowei Ye, Ning Xu, Xiaofeng Liu, Xiao Yao, A. Jiang","doi":"10.1109/CISCE50729.2020.00058","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00058","url":null,"abstract":"The powerful capabilities of deep convolutional neural networks make them useful in various fields. However, most edge devices are difficult to afford the huge amount of parameters and high computational cost. Therefore, it is highly imperative to compress these huge models to make them lightweight to enable real-time inference on edge devices. Channel pruning is a mainstream method of network compression. Generally, the Lasso constraint is imposed on the scaling factor in the batch normalization layer to make them tend to zero for selecting unimportant channels and then prune them. However, Lasso is a non-smooth function that is not derivable at zero, we experimentally find that when the value of the loss function is small, it is difficult to decline continuously. Aiming at the above problems, this paper proposes a pruning strategy based on the derivable function Smooth-Lasso, using Smooth-Lasso as a regularization constraint to perform sparse training and then prune the network. Experiments on benchmark datasets and convolutional networks show that our method can not only make the loss function converge quickly, but also save more storage space and computational cost than the baseline method while maintaining the same level of accuracy as the original network.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116315615","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/CISCE50729.2020.00020
Chao Duan, Dongpo Guo, Jerry Xie, Jing Zhang
ICT advances in the pace of Moore’s law as well as Gilder’s Law. The introduction of ICT into education has resulted astounding effects in learning. This paper presents a case study of the establishment of National Engineering Research Center for E-learning, as part of education informationized effort in China. There are four stages of China’s integration of ICT with education: start-up phase, application phase, integration phase, and innovation phase. The four phases are well in line with the ICT evolutions from standalone computer, to all IP network, to all cloud infrastructure, and to today’s all AI, which utilizes full spectrum of ICT including emerging technology advances in big data and artificial intelligence. The role of ICT in education has gone far beyond as an assisting tool in learning; it led to reconsider the traditional classroom centric teaching architecturally, as well as represented a paradigm shift in education including the roles of classroom teaching, the way of learning, the learning period systematically. Interactive and exploratory learning have become the mantra and education as a service is inevitable. At the same time, like any other technologies, ICT used in education can backfire if not carefully planned. This paper proposes a systematical engineering approach that integrates multiple tooling and activities around the learning of a particular subject with the purpose that engages learners through an intuitive, game-like environment where students learn through exploration and discovery.
{"title":"The Impact of ICT Advances on Education: A Case Study","authors":"Chao Duan, Dongpo Guo, Jerry Xie, Jing Zhang","doi":"10.1109/CISCE50729.2020.00020","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00020","url":null,"abstract":"ICT advances in the pace of Moore’s law as well as Gilder’s Law. The introduction of ICT into education has resulted astounding effects in learning. This paper presents a case study of the establishment of National Engineering Research Center for E-learning, as part of education informationized effort in China. There are four stages of China’s integration of ICT with education: start-up phase, application phase, integration phase, and innovation phase. The four phases are well in line with the ICT evolutions from standalone computer, to all IP network, to all cloud infrastructure, and to today’s all AI, which utilizes full spectrum of ICT including emerging technology advances in big data and artificial intelligence. The role of ICT in education has gone far beyond as an assisting tool in learning; it led to reconsider the traditional classroom centric teaching architecturally, as well as represented a paradigm shift in education including the roles of classroom teaching, the way of learning, the learning period systematically. Interactive and exploratory learning have become the mantra and education as a service is inevitable. At the same time, like any other technologies, ICT used in education can backfire if not carefully planned. This paper proposes a systematical engineering approach that integrates multiple tooling and activities around the learning of a particular subject with the purpose that engages learners through an intuitive, game-like environment where students learn through exploration and discovery.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"38 1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123207724","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/CISCE50729.2020.00084
Chao Liang, Shaojie Xin
With the continuous innovation and development of artificial intelligence, the theoretical research on and application of deep learning, one of its branches, has also reached a certain height, and has become a research hotspot in all walks of life. In the medical field, traditional manual image reading and other medical image analysis methods have been unable to adapt to the sharp increase in the amount of impact data. Based on this, the combination of deep learning and medical imaging has eased this pressure. This article first briefly analyzes the relevant theories of deep learning, and focuses on its applications in medical image classification and recognition, medical image segmentation, and computer-aided diagnosis. Finally, the application of deep learning in medical images is prospected.
{"title":"Research Status and Prospects of Deep Learning in Medical Images","authors":"Chao Liang, Shaojie Xin","doi":"10.1109/CISCE50729.2020.00084","DOIUrl":"https://doi.org/10.1109/CISCE50729.2020.00084","url":null,"abstract":"With the continuous innovation and development of artificial intelligence, the theoretical research on and application of deep learning, one of its branches, has also reached a certain height, and has become a research hotspot in all walks of life. In the medical field, traditional manual image reading and other medical image analysis methods have been unable to adapt to the sharp increase in the amount of impact data. Based on this, the combination of deep learning and medical imaging has eased this pressure. This article first briefly analyzes the relevant theories of deep learning, and focuses on its applications in medical image classification and recognition, medical image segmentation, and computer-aided diagnosis. Finally, the application of deep learning in medical images is prospected.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128598310","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2020-07-01DOI: 10.1109/cisce50729.2020.00047
Wu Jing
Currently, people need to use a large number of accounts and password, this has already become a problem which can be called "sea of accounts and passwords". This problem has a lot of negative effects. This paper presents a new kind of user identity credential, i.e. user activity map. On this basis, a decentralized user identity authentication model based on activity proof is proposed. Based on the blockchain idea, the model is built by Authentication Chain. The Authentication Chain is composed of Activity Chain and Activity Proofer Chain. This model does not need a central system, but can use the user’s activity map as the identity credential, and verify the user’s identity with the supporting results of activity map through multiple systems. This paper introduces the prototype verification of the model, and discusses the security and feasibility of the model.
{"title":"A Decentralized User Authentication Model Based on Activity Proof : Use the new user identity credential: activity map","authors":"Wu Jing","doi":"10.1109/cisce50729.2020.00047","DOIUrl":"https://doi.org/10.1109/cisce50729.2020.00047","url":null,"abstract":"Currently, people need to use a large number of accounts and password, this has already become a problem which can be called \"sea of accounts and passwords\". This problem has a lot of negative effects. This paper presents a new kind of user identity credential, i.e. user activity map. On this basis, a decentralized user identity authentication model based on activity proof is proposed. Based on the blockchain idea, the model is built by Authentication Chain. The Authentication Chain is composed of Activity Chain and Activity Proofer Chain. This model does not need a central system, but can use the user’s activity map as the identity credential, and verify the user’s identity with the supporting results of activity map through multiple systems. This paper introduces the prototype verification of the model, and discusses the security and feasibility of the model.","PeriodicalId":101777,"journal":{"name":"2020 International Conference on Communications, Information System and Computer Engineering (CISCE)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121491011","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：