Pub Date : 2020-12-11DOI: 10.1109/TOCS50858.2020.9339755
Shen Jianjie, Zou Weijun
Violence detection based on deep learning is a research hotspot in intelligent video surveillance. The pre-trained Three-Dimensional convolutional network (C3D) has a weak ability to extract spatiotemporal features of video. It can only achieve an accuracy of 88.2% on the UCF-101 data set, which cannot meet the accuracy requirements for detecting violent behavior in videos. Thus, this paper proposes a network architecture based on the C3D and fusion of the Inception-Resnet-v2 network residual Inception module. Through adaptive learning of feature weights, the three-dimensional features of violent behavior videos can be fully explored and the ability to express features is enhanced. Secondly, in view of the small amount of data in the data set for violence detection (HockeyFights), which easily leads to the problems of overfitting and low generalization ability, the UCF101 data set is used for fine-tune, so that the shallow layer of the network can fully extract the spatiotemporal features; Finally, the use of quantization tools to quantify network parameters and adjusting the sliding window parameters during inference can effectively improves the inference efficiency and improves the real-time performance while ensuring high accuracy. Through experiments, the accuracy of the network designed in the paper on the UCF-101 dataset is improved by 6.1% compared to the C3D network, and by 3.1% compared with the R3D network, indicating that the improved network structure can extract more spatiotemporal features, and finally achieved an accuracy of 95.1% on the HockeyFights test set.
{"title":"Violence Detection Based on Three-Dimensional Convolutional Neural Network with Inception-ResNet","authors":"Shen Jianjie, Zou Weijun","doi":"10.1109/TOCS50858.2020.9339755","DOIUrl":"https://doi.org/10.1109/TOCS50858.2020.9339755","url":null,"abstract":"Violence detection based on deep learning is a research hotspot in intelligent video surveillance. The pre-trained Three-Dimensional convolutional network (C3D) has a weak ability to extract spatiotemporal features of video. It can only achieve an accuracy of 88.2% on the UCF-101 data set, which cannot meet the accuracy requirements for detecting violent behavior in videos. Thus, this paper proposes a network architecture based on the C3D and fusion of the Inception-Resnet-v2 network residual Inception module. Through adaptive learning of feature weights, the three-dimensional features of violent behavior videos can be fully explored and the ability to express features is enhanced. Secondly, in view of the small amount of data in the data set for violence detection (HockeyFights), which easily leads to the problems of overfitting and low generalization ability, the UCF101 data set is used for fine-tune, so that the shallow layer of the network can fully extract the spatiotemporal features; Finally, the use of quantization tools to quantify network parameters and adjusting the sliding window parameters during inference can effectively improves the inference efficiency and improves the real-time performance while ensuring high accuracy. Through experiments, the accuracy of the network designed in the paper on the UCF-101 dataset is improved by 6.1% compared to the C3D network, and by 3.1% compared with the R3D network, indicating that the improved network structure can extract more spatiotemporal features, and finally achieved an accuracy of 95.1% on the HockeyFights test set.","PeriodicalId":373862,"journal":{"name":"2020 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)","volume":"114 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123373186","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 : 2020-12-11DOI: 10.1109/TOCS50858.2020.9339747
Xinge Li, Jiaying Zhang, W. Jin, Weili Liu
As one of the key applications of the Internet of things technology, the development of intelligent breeding system has completely changed the traditional artificial feeding mode of poultry, used modern means to remote real-time monitor the environment of the breeding place, liberated the breeding personnel from the traditional heavy breeding work, and greatly improved the work efficiency. In this paper, the key technologies of 5g intelligent IoT poultry breeding system are studied. Taking poultry breeding as the research object, an intelligent poultry breeding management system is constructed. On the basis of solving the storage and processing problems of massive perceptual data, the real-time collection and query of breeding process data, real-time monitoring of poultry growth status and poultry house environment, as well as the digital production and operation management of farms are realized, which provides scientific basis for poultry breeding managers to query, manage and make decisions on poultry information. A remote monitoring system of poultry breeding environment based on Internet of things is designed, which is convenient for poultry production managers to obtain and monitor the breeding environment information anytime and anywhere, without the limitation of time and place. Undoubtedly, the development of the system has important practical significance for the informatization and intelligent management of poultry production.
{"title":"Key Technology Implementation of Poultry Breeding System for 5G Intelligent IOT","authors":"Xinge Li, Jiaying Zhang, W. Jin, Weili Liu","doi":"10.1109/TOCS50858.2020.9339747","DOIUrl":"https://doi.org/10.1109/TOCS50858.2020.9339747","url":null,"abstract":"As one of the key applications of the Internet of things technology, the development of intelligent breeding system has completely changed the traditional artificial feeding mode of poultry, used modern means to remote real-time monitor the environment of the breeding place, liberated the breeding personnel from the traditional heavy breeding work, and greatly improved the work efficiency. In this paper, the key technologies of 5g intelligent IoT poultry breeding system are studied. Taking poultry breeding as the research object, an intelligent poultry breeding management system is constructed. On the basis of solving the storage and processing problems of massive perceptual data, the real-time collection and query of breeding process data, real-time monitoring of poultry growth status and poultry house environment, as well as the digital production and operation management of farms are realized, which provides scientific basis for poultry breeding managers to query, manage and make decisions on poultry information. A remote monitoring system of poultry breeding environment based on Internet of things is designed, which is convenient for poultry production managers to obtain and monitor the breeding environment information anytime and anywhere, without the limitation of time and place. Undoubtedly, the development of the system has important practical significance for the informatization and intelligent management of poultry production.","PeriodicalId":373862,"journal":{"name":"2020 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126859013","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 : 2020-12-11DOI: 10.1109/TOCS50858.2020.9339760
Lei Tong
The paper discusses the significance of building a smart transportation system, analyzes the system structure, and focuses on the overall framework, system functions, database structure and optimal path analysis methods of the smart transportation system. Using the deployed Hadoop server cluster, using Map The improved Apriori algorithm of the Reduce programming model, the instance runs traffic big data, carries on the traffic flow analysis and the speed and overspeed analysis, and digs out the information that is beneficial to traffic control. The system integrates a variety of modern technologies such as communication technology and control technology, and effectively combines the communication between people, vehicles and roads. At the same time, the system proved the feasibility and effectiveness of using the Hadoop platform to mine massive traffic information.
{"title":"Research on the Application of Key Technologies in the Construction of Smart Cities Based on Smart Transportation","authors":"Lei Tong","doi":"10.1109/TOCS50858.2020.9339760","DOIUrl":"https://doi.org/10.1109/TOCS50858.2020.9339760","url":null,"abstract":"The paper discusses the significance of building a smart transportation system, analyzes the system structure, and focuses on the overall framework, system functions, database structure and optimal path analysis methods of the smart transportation system. Using the deployed Hadoop server cluster, using Map The improved Apriori algorithm of the Reduce programming model, the instance runs traffic big data, carries on the traffic flow analysis and the speed and overspeed analysis, and digs out the information that is beneficial to traffic control. The system integrates a variety of modern technologies such as communication technology and control technology, and effectively combines the communication between people, vehicles and roads. At the same time, the system proved the feasibility and effectiveness of using the Hadoop platform to mine massive traffic information.","PeriodicalId":373862,"journal":{"name":"2020 IEEE Conference on Telecommunications, Optics and Computer Science (TOCS)","volume":"1721 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-12-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129436109","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：