View Item 
      •   IPB Repository
      • Dissertations and Theses
      • Undergraduate Theses
      • UT - Faculty of Mathematics and Natural Sciences
      • UT - Computer Science
      • View Item
      •   IPB Repository
      • Dissertations and Theses
      • Undergraduate Theses
      • UT - Faculty of Mathematics and Natural Sciences
      • UT - Computer Science
      • View Item
      JavaScript is disabled for your browser. Some features of this site may not work without it.

      Pembangunan Model Deep Learning untuk Mengidentifikasi Tanaman Kelapa Sawit melalui Citra Drone

      Thumbnail
      View/Open
      Cover (369.4Kb)
      Fulltext (5.191Mb)
      Lampiran (3.537Mb)
      Date
      2023
      Author
      Tachtiar, Laudza Muhammad Afin
      Mushthofa, Mushthofa
      Metadata
      Show full item record
      Abstract
      Komoditas kelapa sawit merupakan salah satu keunggulan dari subsektor perkebunan Indonesia. Pengindraan jauh dapat diadaptasi pada perkebunan kelapa sawit adalah dengan melakukan upaya perhitungan jumlah pohon kelapa sawit pada suatu cakupan area tertentu. Penghitungan jumlah pohon kelapa sawit merupakan praktik perkebunan yang dapat berguna untuk inventarisasi aset biologis, estimasi jumlah produksi kelapa sawit, dan pemantauan pertumbuhan tanaman. Penelitian ini bertujuan untuk menerapkan metode YOLOv8s dan YOLOv8m untuk sistem identifikasi dan perhitungan pohon kelapa sawit secara otomatis berdasarkan tangkapan citra drone, lalu membandingkan hasilnya dengan penelitian sebelumnya yang menggunakan algoritme YOLOv3, YOLOv4, dan YOLOv5m. Dari delapan model yang didapat dari penelitian ini, diperoleh bahwa model terbaik untuk YOLOv8s adalah model dengan input size 832 dan batch 16 dengan F1-score sebesar 0,996 dan mAP50-95 sebesar 0,995. Sedangkan pada YOLOv8m adalah model dengan input size 608 dan batch 32 dengan hasil F1-score sebesar 0,996 dan mAP50-95 sebesar 0,995. Delapan model baru yang diperoleh relatif memiliki performa yang lebih baik dari penelitian sebelumnya.
       
      The palm oil commodity is one of the advantages of the Indonesian plantation sub-sector. Remote sensing that can be adapted to oil palm plantations is to make efforts to calculate the number of oil palm trees in a certain coverage area. Calculating the number of oil palm trees is a plantation practice that can be useful for inventorying biological assets, estimating the amount of oil palm, and monitoring plant growth. This study aims to apply the YOLOv8s and YOLOv8m methods to an automatic assistance and calculation system for oil palm trees based on drone image capture, and then compare the results with previous studies using the YOLOv3, YOLOv4, and YOLOv5m algorithms. The eight models which obtained from this study, it was found that the best model for YOLOv8s was the model with an input size of 832 and a batch of 16 with an F1-score of 0.996 and a mAP50-95 of 0.995. Whereas the YOLOv8m is a model with an input size of 608 and a batch of 32 with an F1-score of 0.996 and a mAP50-95 of 0.995. The eight new models obtained have relatively better performance than previous studies.
       
      URI
      http://repository.ipb.ac.id/handle/123456789/124672
      Collections
      • UT - Computer Science [2482]

      Copyright © 2020 Library of IPB University
      All rights reserved
      Contact Us | Send Feedback
      Indonesia DSpace Group 
      IPB University Scientific Repository
      UIN Syarif Hidayatullah Institutional Repository
      Universitas Jember Digital Repository
        

       

      Browse

      All of IPB RepositoryCollectionsBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

      My Account

      Login

      Application

      google store

      Copyright © 2020 Library of IPB University
      All rights reserved
      Contact Us | Send Feedback
      Indonesia DSpace Group 
      IPB University Scientific Repository
      UIN Syarif Hidayatullah Institutional Repository
      Universitas Jember Digital Repository