Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/170194
Full metadata record
DC FieldValueLanguage
dc.contributor.advisorSupriyanto-
dc.contributor.authoralfarizi, Farhan ali-
dc.date.accessioned2025-08-23T08:18:07Z-
dc.date.available2025-08-23T08:18:07Z-
dc.date.issued2025-
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/170194-
dc.description.abstractPemantauan kondisi tanaman secara spasial menjadi aspek penting dalam pengelolaan kebun buah, termasuk tanaman jambu biji (Psidium guajava L). Citra multispektral dari drone menyediakan data resolusi tinggi yang mampu merekam perbedaan spektral vegetasi secara detail. Namun, segmentasi objek kanopi secara akurat masih menjadi tantangan, terutama dalam membedakan kanopi pohon dari latar belakang seperti tanah, rumput, dan belukar. Penelitian ini merancang model segmentasi berbasis deep learning dengan arsitektur U-Net untuk memisahkan kanopi jambu biji dari latar belakang menggunakan input citra multispektral drone yang telah diproses melalui band stacking dan perhitungan indeks vegetasi. Model dilatih menggunakan citra patch berukuran 256 × 256 piksel dan diuji performanya menggunakan metrik evaluasi seperti Precision, Recall, F1-Score, dan Intersection over Union (IoU). Hasil menunjukkan bahwa model mampu mengenali objek kanopi dengan akurasi yang tinggi. Temuan ini mengindikasikan bahwa model yang dirancang efektif dalam segmentasi spasial objek vegetasi, serta dapat beroperasi dengan baik dalam memisahkan kanopi pohon jambu biji dari tutupan lahan lain seperti tanah, rumput, dan belukar.-
dc.description.abstractSpatial monitoring of crop conditions is a crucial aspect in managing fruit orchards, including guava (Psidium guajava L.) plantations. Multispectral imagery from Unmanned Aerial Vehicles (UAV) or Drone, provides high-resolution data capable of capturing detailed spectral differences in vegetation. Segmenting canopy objects remains a challenge, particularly in distinguishing tree canopies from backgrounds such as soil, grass, and shrubs. This study proposes a deep learningbased segmentation model using the U-Net architecture to separate guava canopies from the background by utilizing drone multispectral imagery that has been preprocessed through band stacking and vegetation index calculation. The model was trained on 256×256 pixel image patches and evaluated using metrics such as Precision, Recall, F1-Score, and Intersection over Union (IoU). The results show that the model is capable of recognizing canopy objects with high accuracy. This finding indicates that the designed model is effective in spatial segmentation of vegetation objects and can perform well in separating guava tree canopies from other land cover such as soil, grass, and shrubs-
dc.description.sponsorshipAgribusiness and Technology Park IPB (ATP IPB)-
dc.language.isoid-
dc.publisherIPB Universityid
dc.titlePERANCANGAN MODEL DETEKSI KANOPI JAMBU BIJI (Psidium guajava L.) BERBASIS SEGMENTASI CITRA DRONE MULTISPEKTRAL DAN DEEP LEARNINGid
dc.title.alternativeDEVELOPMENT CANOPY DETECTION MODEL OF GUAVA (psidium guajava L.) TREES USING MULTISPECTRAL DRONE IMAGE SEGMENTATION AND DEEP LEARNING-
dc.typeSkripsi-
dc.subject.keywordDrone Multispektralid
dc.subject.keywordjambu bijiid
dc.subject.keyworddeep learningid
dc.subject.keywordsegmentasi kanopiid
dc.subject.keywordU-Netid
dc.subject.keywordmultispectral droneid
dc.subject.keywordsegmentation of guava canopyid
Appears in Collections:UT - Agricultural and Biosystem Engineering

Files in This Item:
File Description SizeFormat 
cover_F1401211068_53e39824af824d9bbbdea2d8e38f7bca.pdfCover449.48 kBAdobe PDFView/Open
fulltext_F1401211068_7b68dc7c32e542a492e12e6c52014831.pdf
  Restricted Access
Fulltext6.8 MBAdobe PDFView/Open
lampiran_F1401211068_0aa9ca990ae9418d8c90cf018e55965b.pdf
  Restricted Access
Lampiran308.71 kBAdobe PDFView/Open


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