Please use this identifier to cite or link to this item:
http://repository.ipb.ac.id/handle/123456789/162617| Title: | Identifikasi Varietas Jagung Menggunakan Metode Convolutional Neural Network |
| Other Titles: | Identification of Corn Varieties Using the Convolutional Neural Network Method |
| Authors: | Kustiyo, Aziz Idin, Abdurrahim Ramadhan |
| Issue Date: | 2025 |
| Publisher: | IPB University |
| Abstract: | Jagung merupakan salah satu komoditas utama di Indonesia, namun
produktivitasnya sangat dipengaruhi oleh kondisi lingkungan dan cuaca. Fluktuasi
hasil panen jagung sering terjadi akibat perubahan iklim, seperti fenomena El
Niño yang memicu kekeringan dan keterlambatan musim tanam. Oleh karena itu,
pemilihan benih unggul yang tahan terhadap cekaman kekeringan menjadi sangat
penting, terutama di daerah dengan kondisi lahan kering seperti Nusa Tenggara
Timur. Metode identifikasi benih yang masih menggunakan pendekatan
tradisional, seperti pengukuran morfologi secara manual, dianggap kurang efisien
baik dari segi waktu maupun tenaga. Penelitian ini mengembangkan model
klasifikasi varietas jagung menggunakan Convolutional Neural Network (CNN)
dengan objek penelitian lima varietas yaitu NK-212, NK-7328, P-21, Pertiwi-2,
dan Pertiwi-6. Tiga arsitektur CNN dengan tingkat kompleksitas berbeda
dirancang dan diuji. Hasilnya, model terbaik yang terdiri dari tiga lapisan
konvolusi dan dua lapisan dense dengan dropout menunjukkan performa optimal,
mencapai akurasi 89,20% pada data pengujian. Corn is one of the main commodities in Indonesia, but its productivity is greatly influenced by environmental and weather conditions. Fluctuations in corn harvests often occur due to climate change, such as the El Niño phenomenon which triggers drought and delays in the planting season. Therefore, the selection of superior seeds that are resistant to drought stress is very important, especially in areas with dry land conditions such as East Nusa Tenggara. Seed identification methods that still use traditional approaches, such as manual morphological measurements, are considered less efficient in terms of time and energy. This study developed a corn variety classification model using Convolutional Neural Network (CNN) with five varieties as research objects, namely NK-212, NK-7328, P-21, Pertiwi-2, and Pertiwi-6. Three CNN architectures with different levels of complexity were designed and tested. As a result, the best model consisting of three convolutional layers and two dense layers with dropout showed optimal performance, achieving an accuracy of 89.20% on the test data. |
| URI: | http://repository.ipb.ac.id/handle/123456789/162617 |
| Appears in Collections: | UT - Computer Science |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| cover_G6401211114_bcbd318549f946a0937287d222ff3a9f.pdf | Cover | 884.9 kB | Adobe PDF | View/Open |
| fulltext_G6401211114_720820c82f3b4cfd8dd0a693f71c45c8.pdf Restricted Access | Fulltext | 1.63 MB | Adobe PDF | View/Open |
| lampiran_G6401211114_2cee999bd7594d2a8103416d7379f7f6.pdf Restricted Access | Lampiran | 1.89 MB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.