Please use this identifier to cite or link to this item: http://dspace.ensta.edu.dz/jspui/handle/123456789/580
Full metadata record
DC FieldValueLanguage
dc.contributor.authorBENYOUCEF, Meriem-
dc.contributor.authorCHOUAF, Seloua (Directeur de thèse)-
dc.date.accessioned2026-11-11T08:08:51Z-
dc.date.available2026-11-11T08:08:51Z-
dc.date.issued2022-
dc.identifier.urihttp://dspace.ensta.edu.dz/jspui/handle/123456789/580-
dc.descriptionMémoire de fin d’étude de Master : Systèmes de Télécommunications Et Réseaux: Alger, Ecole Nationale Supérieure des Technologie Avancées (ex ENST): 2022en_US
dc.description.abstractThe process of generating an accurate textual description based on a given image utilizing deep learning-based architectures has been applied in many fields. In this paper, we chose to develop and compare the performances of multimodal CNN and LSTM based architectures applied for the remote sensing task. First, we aimed to evaluate our architectures on the RSCID (Remote Sensing Image Captioning Dataset) in order to compare our results with what was achieved in literature. Second, we studied the performances of our architectures on a remote sensing dataset of our creation. Our various experiments demonstrated the efficiency of different CNN and LSTM multimodal architectures in describing remote sensing images accurately scoring satisfactory results in terms of natural language metrics.en_US
dc.language.isoenen_US
dc.publisherENSTen_US
dc.subjectRemote sensingen_US
dc.subjectConvolutional Neural Networks CNNen_US
dc.subjectRecurrent Neural Networks RNNen_US
dc.subjectNatural Language Processing NLPen_US
dc.subjectNatural Language Processing NLPen_US
dc.subjectLong Short Term MemoryLSTMen_US
dc.subjectdeep learningen_US
dc.titlePerformance Evaluation Of An Automated Image Description Routine Considering Different Datasetsen_US
dc.typeThesisen_US
Appears in Collections:MFE- Systèmes de Télécommunications et Réseaux

Files in This Item:
File Description SizeFormat 
master.2022.STR.BENYOUCEF.Meriem.pdfMémoire de Master1.74 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.