Nuqqy Zahhar, Ahmad Haikal (2026) KLASIFIKASI JENIS DURIAN BERDASARKAN CITRA DAUN MENGGUNAKAN METODE CONVOLUTIONAL NEURAL NETWORK DENGAN ARSITEKTUR VGG-16. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
COVER_removed.pdf Download (784kB) |
|
Text (BAB 1)
BAB I.pdf Download (296kB) |
|
Text (BAB 2)
BAB II.pdf Restricted to Repository staff only until 5 March 2028. Download (544kB) |
|
Text (BAB 3)
BAB III.pdf Restricted to Repository staff only until 5 March 2028. Download (513kB) |
|
Text (BAB 4)
BAB IV.pdf Restricted to Repository staff only until 5 March 2028. Download (2MB) |
|
Text (BAB 5)
BAB V.pdf Download (231kB) |
|
Text (Daftar Pustaka)
Daftar Pustaka.pdf Download (213kB) |
|
Text (Lampiran)
Lampiran.pdf Restricted to Repository staff only until 5 March 2028. Download (194kB) |
Abstract
Durian (Durio zibethinus Murr) is a genetic asset of Indonesia's local plants with high economic value. The quality of local durian still faces challenges such as inadequate maintenance, resulting in a decline in sales value. This condition arises from farmers' limited understanding in determining the most suitable durian varieties for cultivation. Accurate identification of durian species is crucial in increasing the quality and economic value of durian fruit. Taxonomic research indicates that leaf morphological characteristics can be a reliable taxonomic indicator for plant species identification. Manually identifying durian species based on leaves requires specialized expertise and considerable time, and is susceptible to observer subjectivity. In Nganjuk Regency, there are three popular durian varieties: local durian, lai durian, and montong durian, which share similar leaf characteristics, making them difficult for farmers to visually distinguish. Previous studies have not specifically examined durian species classification based on leaf images using deep learning. Therefore, there is a research gap that needs to be explored to support the development of precision agriculture in Indonesia. This study applies the CNN method using the VGG-16 architecture to classify durian types based on leaf images. Durian types are divided into 3, namely: montong, lokal and Lai. The dataset used was 600, with a balanced distribution of 200 images per class. The obtained dataset was preprocessed and augmented to increase the number of datasets by 4x to 2400. Next, testing was carried out in 16 model scenarios. With 4 different dataset splits and 4 optimizers: SGD, Adam, Adamax and Adagard. From 16 test scenarios, the best model was obtained with an accuracy of 97.08%. The data split was 90:10 and the optimizer used was adam. These results demonstrate that the system can classify durian types based on leaf images well and with high accuracy
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | T Technology > T Technology (General) > T385 Computer Graphics T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5105 Computer Network T Technology > TP Chemical technology > TP368 Food processing and manufacture |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Ahmad Haikal Nuqqy Zahhar | ||||||||||||
| Date Deposited: | 13 Mar 2026 01:56 | ||||||||||||
| Last Modified: | 13 Mar 2026 01:56 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/50207 |
Actions (login required)
 |
View Item |