Please use this identifier to cite or link to this item:
https://dspace.univ-ouargla.dz/jspui/handle/123456789/34075
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.author | Amran, Laila | - |
dc.contributor.author | Ferhi, Aya | - |
dc.date.accessioned | 2023-09-17T10:06:43Z | - |
dc.date.available | 2023-09-17T10:06:43Z | - |
dc.date.issued | 2023 | - |
dc.identifier.uri | https://dspace.univ-ouargla.dz/jspui/handle/123456789/34075 | - |
dc.description.abstract | Soil plays an important role in the quality of agricultural crops, especially since we are in an era in which the agricultural sector occupies a great importance not only from the side of the econ omy, but also from the side of achieving food sufficiency, so if the type of the soil is not appropri ate, then the product will not be of the required quality. In our research, we wanted to help every farmer and peasant and everyone interested in this field by developing a website that allows soil classification (initially five types) by including a picture of the soil to know its type. The classifi cation process was carried out using deep learning exactly Proposed CNN model, which had the highest accuracy(86%) after comparing it with 3 other models, VGG(80%), ResNet(79%), and MobileNet(76%). In the end, after the classification is done, the most important characteristics (six characteristics) of that soil are presented, whitch is temperature, Ph,Porosity ,texture,color ,density | en_US |
dc.language.iso | en | en_US |
dc.publisher | University Kasdi Merbah– OUARGLA | en_US |
dc.subject | Artificial Intelligence | en_US |
dc.subject | Deep Learning | en_US |
dc.subject | CNN | en_US |
dc.subject | Proposed CNN | en_US |
dc.subject | VGG16 | en_US |
dc.subject | Res Net | en_US |
dc.subject | Mobile Net | en_US |
dc.title | Soil recognition and features measurement using AI | en_US |
dc.type | Thesis | en_US |
Appears in Collections: | Département d'informatique et technologie de l'information - Master |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.