Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributorTEJENDRA KUMAR YADAVen
dc.contributorTejendra Kumar Yadavth
dc.contributor.advisorNattapon Mahaviken
dc.contributor.advisorนัฐพล มหาวิคth
dc.contributor.otherNaresuan University. Faculty of Agriculture,Natural Resources and Environmenten
dc.descriptionMaster of Science (M.S.)en
dc.descriptionวิทยาศาสตรมหาบัณฑิต (วท.ม.)th
dc.description.abstractMine reclamation is essential as the extractive organizations are bounded by-laws that have been established by stakeholders to ensure the restoration of mined areas. This study focuses on the application of Unmanned Aerial Vehicle (UAV) photogrammetry and deep learning. The use of a UAV improves safety compared to other surveying systems in mines. In addition, monitoring utilizing drones delivers fast, real-time results, and minimizes human exposure in unsafe ground conditions at mines area. UAVs have revitalized the mining industry through Artificial Intelligence. Deep learning reignited the pursuit of artificial intelligence towards machine to perform related tasks in an automated way. The recent advances of Deep learning (DL) for computer vision tasks, especially for Convolution Neural Network (CNN) models, the potentials the automatically classification of land cover using UAV Photogrammetry. Orthophoto and Digital Surface Model (DSM) are the photogrammetric results used for land cover classification. This research aims to employ the classification of land cover for monitoring mine reclamation using DL from the UAV photogrammetric results (orthophoto and DSM) at Mae Moh mine in Lampang, Thailand. Two vegetation areas were selected (Pattern and complex) to perform the classification using DL with CNN also, the height of the trees was calculated using results from UAV photogrammetry.  The land cover was classified into five classes, comprising: 1) trees, 2) shadow, 3) grassland, 4) barren land, and 5) others. The effectiveness of both datasets was examined to verify whether orthophoto or combination of orthophoto with DSM for land cover classification. Land cover classes were, thus, classified with accuracy. The experimental findings revealed that effective results for land cover classification over test area were obtained by DL through the combination of orthophoto and DSM with an Overall Accuracy (OA) of 0.904, Average Accuracy (AA) of 0.681, and Kappa (K) of 0.937 for study area 1 and OA of 0.751, AA of 0.636, K of 0.684 for study area 2. Our experiments presented that land cover classification from combination orthophoto with DSM was more precise than using orthophoto only. This research provides framework for conducting an analytical process, a UAV approach with DL based evaluation of mine reclamation with safety, also providing a time series information for future efforts to evaluate reclamation.  en
dc.publisherNaresuan Universityen_US
dc.rightsNaresuan Universityen_US
dc.subjectUAV photogrammetry, Land cover classification, Deep learning (DL), Convolution neural network (CNN)th
dc.subjectUAV photogrammetry Land cover classification Deep learning (DL) Convolution neural network (CNN)en
dc.subject.classificationEnvironmental Scienceen
dc.title Development of monitoring open-pit mine rehabilitation based on UAV photogrammetry and deep learningen
Appears in Collections:คณะเกษตรศาสตร์ ทรัพยากรธรรมชาติและสิ่งแวดล้อม

Files in This Item:
File Description SizeFormat 
62063188.pdf3.62 MBAdobe PDFView/Open

Items in NU Digital Repository are protected by copyright, with all rights reserved, unless otherwise indicated.