The enormous potential of Information and Communication Technologies (ICT) for addressing critical educational issues is generally acknowledged, but its use in the assessment of the complex skills of reading and understanding a text has been very limited to date. The paper contrasts traditional reading assessment protocols with ReadLet, an ICT platform with a tablet front-end, designed to support online monitoring of silent and oral reading abilities in early graders. ReadLet makes use of cloud computing and mobile technology for large-scale data collection and allows the time alignment of the child's reading behaviour with texts tagged using Natural Language Processing (NLP) tools. Initial findings replicate established benchmarks from the psycholinguistic literature on reading in both typically and atypically developing children, making the application a new ground-breaking approach in the evaluation of reading skills.
{"title":"Using mobile technology for reading assessment","authors":"Loukia Taxitari, Claudia Cappa, Marcello Ferro, Claudia Marzi, Andrea Nadalini, Vito Pirrelli","doi":"10.1109/CiSt49399.2021.9357173","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357173","url":null,"abstract":"The enormous potential of Information and Communication Technologies (ICT) for addressing critical educational issues is generally acknowledged, but its use in the assessment of the complex skills of reading and understanding a text has been very limited to date. The paper contrasts traditional reading assessment protocols with ReadLet, an ICT platform with a tablet front-end, designed to support online monitoring of silent and oral reading abilities in early graders. ReadLet makes use of cloud computing and mobile technology for large-scale data collection and allows the time alignment of the child's reading behaviour with texts tagged using Natural Language Processing (NLP) tools. Initial findings replicate established benchmarks from the psycholinguistic literature on reading in both typically and atypically developing children, making the application a new ground-breaking approach in the evaluation of reading skills.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129138253","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357275
Youness Abouqora, Omar Herouane, L. Moumoun, T. Gadi
Owing to the wide spread of the 3D objects technologies, learning 3D objects labeling and segmentation is becoming one of the most provocative tasks in computer vision and digital multimedia. Recently, convolutional neural network (CNN) has proved itself as a powerful model in segmentation and classification by giving excellent performances. The use of a graphical model such as conditional random field (CRF) contributes further in capturing contextual information and thus improving the segmentation performance. In this paper, we implement a deep architecture in order to segment and label 3D shape parts by considering both spectral and geometric features via a combined framework consisting of a CNN and CRF models. First, low-level features are used to learn deep features using a CNN model, and then formulate the deep CRF model with a CNN-based unary and pairwise potential functions to effectively extract the semantic correlations between adjacent triangles on the mesh. By comparing our results with those from several state-of-the-art, our method shows promising potentials.
{"title":"A Hybrid CNN-CRF Inference Models for 3D Mesh Segmentation","authors":"Youness Abouqora, Omar Herouane, L. Moumoun, T. Gadi","doi":"10.1109/CiSt49399.2021.9357275","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357275","url":null,"abstract":"Owing to the wide spread of the 3D objects technologies, learning 3D objects labeling and segmentation is becoming one of the most provocative tasks in computer vision and digital multimedia. Recently, convolutional neural network (CNN) has proved itself as a powerful model in segmentation and classification by giving excellent performances. The use of a graphical model such as conditional random field (CRF) contributes further in capturing contextual information and thus improving the segmentation performance. In this paper, we implement a deep architecture in order to segment and label 3D shape parts by considering both spectral and geometric features via a combined framework consisting of a CNN and CRF models. First, low-level features are used to learn deep features using a CNN model, and then formulate the deep CRF model with a CNN-based unary and pairwise potential functions to effectively extract the semantic correlations between adjacent triangles on the mesh. By comparing our results with those from several state-of-the-art, our method shows promising potentials.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114184553","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357292
N. Lakhlef, H. Oudira, C. Dumond
The work concerns the problem of monitoring linear antenna arrays using a new scheme denoted as grey wolf optimization (GWO) algorithm. When a strange event (fault) affects an antenna array, the radiation diagram changes and important deviation from the preferred pattern can occur. In this work, re-adjusted of the amplitude and phase distribution of the lasting working elements in a faulty array is considered. This latter can improve the side lobe levels (SLL) and also keep the directivity. The main point of using the GWO technique is its ease of implementation and a high performance computational technique. To assess the strength of this new scheme, different types of failures as case studies were performed. Simulation results evidently have shown the efficiency of the proposed algorithm to correct the failure correction of linear antenna arrays.
{"title":"Failure Correction of Linear Antenna Array using Grey Wolf Optimization","authors":"N. Lakhlef, H. Oudira, C. Dumond","doi":"10.1109/CiSt49399.2021.9357292","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357292","url":null,"abstract":"The work concerns the problem of monitoring linear antenna arrays using a new scheme denoted as grey wolf optimization (GWO) algorithm. When a strange event (fault) affects an antenna array, the radiation diagram changes and important deviation from the preferred pattern can occur. In this work, re-adjusted of the amplitude and phase distribution of the lasting working elements in a faulty array is considered. This latter can improve the side lobe levels (SLL) and also keep the directivity. The main point of using the GWO technique is its ease of implementation and a high performance computational technique. To assess the strength of this new scheme, different types of failures as case studies were performed. Simulation results evidently have shown the efficiency of the proposed algorithm to correct the failure correction of linear antenna arrays.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125317355","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357185
Zina Boussaada, G. Terrasson, A. Llaria, C. Merlo, O. Curea, Paolo Finotto
The aim of the work presented in this paper is the energy management in residential microgrids. The study focuses on Multi Agent System approach based on sensors and actuators network. The system operation requires a large amount of data both for its architectural description and for its real time control. Dealing with this, the fundamental step in the energy management is the organization of data in order to construct the suitable relational database model. The second step is to analyze the system architecture and identify the agents to use for the system. When data are structured and Multi Agent System is modelled, the energy management algorithms are developed and validated taking into account the data availability and the agents properties.
{"title":"Multi Agent Architecture For Smart Building Energy Management","authors":"Zina Boussaada, G. Terrasson, A. Llaria, C. Merlo, O. Curea, Paolo Finotto","doi":"10.1109/CiSt49399.2021.9357185","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357185","url":null,"abstract":"The aim of the work presented in this paper is the energy management in residential microgrids. The study focuses on Multi Agent System approach based on sensors and actuators network. The system operation requires a large amount of data both for its architectural description and for its real time control. Dealing with this, the fundamental step in the energy management is the organization of data in order to construct the suitable relational database model. The second step is to analyze the system architecture and identify the agents to use for the system. When data are structured and Multi Agent System is modelled, the energy management algorithms are developed and validated taking into account the data availability and the agents properties.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"96 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122794451","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357262
Sunna Torge, Waldemar Hahn, R. Jäkel
Automated text analysis as named entity recognition (NER) heavily relies on large amounts of high-quality training data. Transfer learning approaches aim to overcome the problem of lacking domain-specific training data. In this paper, we investigate different transfer learning approaches to recognize unknown domain-specific entities, including the influence on varying training data size. The experiments are based on the revised German SmartData Corpus, and a baseline model, trained on this corpus.
{"title":"Transfer Learning for Domain-Specific Named Entity Recognition in German","authors":"Sunna Torge, Waldemar Hahn, R. Jäkel","doi":"10.1109/CiSt49399.2021.9357262","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357262","url":null,"abstract":"Automated text analysis as named entity recognition (NER) heavily relies on large amounts of high-quality training data. Transfer learning approaches aim to overcome the problem of lacking domain-specific training data. In this paper, we investigate different transfer learning approaches to recognize unknown domain-specific entities, including the influence on varying training data size. The experiments are based on the revised German SmartData Corpus, and a baseline model, trained on this corpus.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121558743","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357239
Anne-Carole Honfoga, M. Dossou, V. Moeyaert
Signal Space Diversity (SSD) technique was adopted in Digital Video Broadcasting-Terrestrial, second generation (DVB-T2) standard ten years ago in order to increase system performance over fading channels. This technique consists in rotated and cyclic Q-delayed $M$-Quadrature Amplitude Modulation (QAM) and has been proposed without taking into account the complexity induced by the demapper. Due to the complexity of the hardware implementation, the features of this technique have not been effectively exploited for network deployment. Therefore, several low complexity algorithms have been proposed during the last decade in the scientific literature in order to reduce the number of metrics and operators used in the demapping process. This paper proposes an exhaustive overview of all the nine DVB-T2 demapping algorithms existing up to now in the literature by comparing them in terms of parameters, performance and percentage of reduction obtained.
{"title":"Comparison of low complexity demapper algorithms applied to rotated constellation in DVB-T2","authors":"Anne-Carole Honfoga, M. Dossou, V. Moeyaert","doi":"10.1109/CiSt49399.2021.9357239","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357239","url":null,"abstract":"Signal Space Diversity (SSD) technique was adopted in Digital Video Broadcasting-Terrestrial, second generation (DVB-T2) standard ten years ago in order to increase system performance over fading channels. This technique consists in rotated and cyclic Q-delayed $M$-Quadrature Amplitude Modulation (QAM) and has been proposed without taking into account the complexity induced by the demapper. Due to the complexity of the hardware implementation, the features of this technique have not been effectively exploited for network deployment. Therefore, several low complexity algorithms have been proposed during the last decade in the scientific literature in order to reduce the number of metrics and operators used in the demapping process. This paper proposes an exhaustive overview of all the nine DVB-T2 demapping algorithms existing up to now in the literature by comparing them in terms of parameters, performance and percentage of reduction obtained.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125649615","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357303
Aicha Khannat, Hanae Sbaï, L. Kjiri
Nowadays, process mining offers key tools for companies to discover and develop continuously their processes. Moreover, many research works focus on the application of process mining techniques to manage configurable process models. To provide configurable process models with high quality and less complexity using process mining techniques, the preparation of event logs before discovering processes is strongly required. Therefore, various approaches have introduced semantics to derive knowledge from process execution traces. Although, this knowledge is used for discovering variability rules or deriving process variants. However, existing works lack the discovery of semantically enriched configurable process models from event logs. Since event logs are usually combined with semantics and domain ontology for optimizing process mining applications, it seems to be important to introduce a new approach of semantically enriched configurable process discovery. It will be helpful during configuration, validation, and enhancement of configurable processes that are extracted from event logs. In this paper, we present a new framework for event logs preprocessing using ontologies. Our aim is to enrich a collection of event logs with concepts from ontologies as a first step towards discovering a semantically enriched configurable process.
{"title":"Event Logs Pre-processing for Configurable Process Discovery: Ontology-Based Approach","authors":"Aicha Khannat, Hanae Sbaï, L. Kjiri","doi":"10.1109/CiSt49399.2021.9357303","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357303","url":null,"abstract":"Nowadays, process mining offers key tools for companies to discover and develop continuously their processes. Moreover, many research works focus on the application of process mining techniques to manage configurable process models. To provide configurable process models with high quality and less complexity using process mining techniques, the preparation of event logs before discovering processes is strongly required. Therefore, various approaches have introduced semantics to derive knowledge from process execution traces. Although, this knowledge is used for discovering variability rules or deriving process variants. However, existing works lack the discovery of semantically enriched configurable process models from event logs. Since event logs are usually combined with semantics and domain ontology for optimizing process mining applications, it seems to be important to introduce a new approach of semantically enriched configurable process discovery. It will be helpful during configuration, validation, and enhancement of configurable processes that are extracted from event logs. In this paper, we present a new framework for event logs preprocessing using ontologies. Our aim is to enrich a collection of event logs with concepts from ontologies as a first step towards discovering a semantically enriched configurable process.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"14 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128879577","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357285
Hanane Zih, Maha el Biadi, Zakriyae Chatri
In recent decades, advancements in information technology have proved essential in a variety of disciplines. Linguistics in general, and Critical Discourse Analysis in particular are no exception, as practitioners working within these areas believe that drawing upon certain computerized methods enables them to obtain accurate results from which they can draw valid generalizations. Relying on computer-assisted text analysis has, for this reason, become a necessity especially among critical discourse analysts and has, as a result of that, led to the proliferation of natural language processing programs geared for the analysis of large linguistic corpuses which researchers interested in the study of language can employ. This variety of programs, however, leads one to wonder about their level of accuracy and effectiveness in identifying the specific linguistic features of interest to the researcher, and their ability to reach the same level of preciseness that a manual analysis of the texts under analysis can achieve. The present paper aims, therefore, to compare two of the mostly used types of corpus linguistics programs to analyze linguistic data in the area of textual analysis to find out about the extent to which there are similarities between them, in terms of the findings they yield. It will draw some comparisons between two programs, namely LancsBox and AntConc, in order to find out about their effectiveness in culling the linguistic features such as passive and nominalized constructions from large amounts of linguistic data. We will after that proceed to comparing them to the results obtained through a manual analysis of the data to see if there are any differences. A corpus of ten news articles published in The Times online newspaper were downloaded and analyzed both manually and digitally so as to examine the occurrences and the distribution of the two aforementioned grammatical constructions in the news reports. The findings show that corpus linguistic software can reliably extract passives and active clauses from the texts. Although both LancsBox and AntConc revealed approximately the same frequencies compared to the findings we obtained manually, both programs did not help in specifically isolating the clauses in which the perpetrator of the action that the articles talk about.
{"title":"Evaluating the effectiveness of corpus linguistic software in analyzing the grammatical structure: LancsBox and AntConc as case studies","authors":"Hanane Zih, Maha el Biadi, Zakriyae Chatri","doi":"10.1109/CiSt49399.2021.9357285","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357285","url":null,"abstract":"In recent decades, advancements in information technology have proved essential in a variety of disciplines. Linguistics in general, and Critical Discourse Analysis in particular are no exception, as practitioners working within these areas believe that drawing upon certain computerized methods enables them to obtain accurate results from which they can draw valid generalizations. Relying on computer-assisted text analysis has, for this reason, become a necessity especially among critical discourse analysts and has, as a result of that, led to the proliferation of natural language processing programs geared for the analysis of large linguistic corpuses which researchers interested in the study of language can employ. This variety of programs, however, leads one to wonder about their level of accuracy and effectiveness in identifying the specific linguistic features of interest to the researcher, and their ability to reach the same level of preciseness that a manual analysis of the texts under analysis can achieve. The present paper aims, therefore, to compare two of the mostly used types of corpus linguistics programs to analyze linguistic data in the area of textual analysis to find out about the extent to which there are similarities between them, in terms of the findings they yield. It will draw some comparisons between two programs, namely LancsBox and AntConc, in order to find out about their effectiveness in culling the linguistic features such as passive and nominalized constructions from large amounts of linguistic data. We will after that proceed to comparing them to the results obtained through a manual analysis of the data to see if there are any differences. A corpus of ten news articles published in The Times online newspaper were downloaded and analyzed both manually and digitally so as to examine the occurrences and the distribution of the two aforementioned grammatical constructions in the news reports. The findings show that corpus linguistic software can reliably extract passives and active clauses from the texts. Although both LancsBox and AntConc revealed approximately the same frequencies compared to the findings we obtained manually, both programs did not help in specifically isolating the clauses in which the perpetrator of the action that the articles talk about.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128224788","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357198
Khalil Saadaoui, K. S. Rhazi, A. Aboudou
Hybrid energies interest many companies and countries. No form of electricity production is optimal in all situations. The wind and the sun are intermittent but do not consume fuel and do not emit greenhouse gases. Natural gas electricity production emits greenhouse gases but is distributable (i.e. it has a yield that can be easily controlled between maximum values of nominal capacity or reduced to zero) to help balance supply and demand. Hydroelectric power often requires and devotes large areas, but is renewable and distributable. However, the realization of all these projects remains dependent on the development of more efficient and more economical electrical energy storage systems. Hybrid power plants: a solution for the future? To provide energy that is more affordable, more reliable, and more sustainable. AI, smart grid [6] and Energy storage: The dynamics of new energies, that is to say, local and renewable, are indeed launched. To succeed in this revolution, the problem of storing renewable energies, due to their intermittent nature, remains to be resolved. Machine learning and neural networks play an important role in improving forecasts in the energy industry.
{"title":"Hybridization and energy storage high efficiency and low cost","authors":"Khalil Saadaoui, K. S. Rhazi, A. Aboudou","doi":"10.1109/CiSt49399.2021.9357198","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357198","url":null,"abstract":"Hybrid energies interest many companies and countries. No form of electricity production is optimal in all situations. The wind and the sun are intermittent but do not consume fuel and do not emit greenhouse gases. Natural gas electricity production emits greenhouse gases but is distributable (i.e. it has a yield that can be easily controlled between maximum values of nominal capacity or reduced to zero) to help balance supply and demand. Hydroelectric power often requires and devotes large areas, but is renewable and distributable. However, the realization of all these projects remains dependent on the development of more efficient and more economical electrical energy storage systems. Hybrid power plants: a solution for the future? To provide energy that is more affordable, more reliable, and more sustainable. AI, smart grid [6] and Energy storage: The dynamics of new energies, that is to say, local and renewable, are indeed launched. To succeed in this revolution, the problem of storing renewable energies, due to their intermittent nature, remains to be resolved. Machine learning and neural networks play an important role in improving forecasts in the energy industry.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"2011 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133209298","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 : 2020-06-05DOI: 10.1109/CiSt49399.2021.9357184
Sara Mobsite, Nabih Alaoui, M. Boulmalf
Elders are one of the huge and fast-growing populations, the majority of them suffers from several daily accidents like fall, and according to the World Health Organization, fall is the second leading cause of accidental injury deaths, for this reason, we proposed a dynamic video classification system in order detect fall with one surveillance camera. Our proposed framework consists of two steps, the first step is the extraction of the human body silhouette from the video frames with the Mask R-CNN, in the second step, we used a combination between the Convolutions Neural Network CNN, and the LSTM long short-time memory to learn long-term dependencies between the successive frames. The best result of our system was achieved when we used 10 frames per video with accuracy, precision, recall, and F1-score equals to 100%.
{"title":"A framework for elders fall detection using deep learning","authors":"Sara Mobsite, Nabih Alaoui, M. Boulmalf","doi":"10.1109/CiSt49399.2021.9357184","DOIUrl":"https://doi.org/10.1109/CiSt49399.2021.9357184","url":null,"abstract":"Elders are one of the huge and fast-growing populations, the majority of them suffers from several daily accidents like fall, and according to the World Health Organization, fall is the second leading cause of accidental injury deaths, for this reason, we proposed a dynamic video classification system in order detect fall with one surveillance camera. Our proposed framework consists of two steps, the first step is the extraction of the human body silhouette from the video frames with the Mask R-CNN, in the second step, we used a combination between the Convolutions Neural Network CNN, and the LSTM long short-time memory to learn long-term dependencies between the successive frames. The best result of our system was achieved when we used 10 frames per video with accuracy, precision, recall, and F1-score equals to 100%.","PeriodicalId":253233,"journal":{"name":"2020 6th IEEE Congress on Information Science and Technology (CiSt)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-06-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124199011","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：