Perbandingan Arsitektur ResNet50 Dan EfficientNetB0 Pada Klasifikasi Penyakit Daun Padi Dengan 5 Optimizer

Huda, Muchammad Syamsu (2025) Perbandingan Arsitektur ResNet50 Dan EfficientNetB0 Pada Klasifikasi Penyakit Daun Padi Dengan 5 Optimizer. Undergraduate thesis, UPN Veteran Jawa Timur.

	Text (Cover) 21081010241.-cover.pdf Download (2MB)
	Text (Bab 1) 21081010241.-bab1.pdf Download (184kB)
	Text (Bab 2) 21081010241.-bab2.pdf Restricted to Repository staff only until 5 December 2028. Download (1MB)
	Text (Bab 3) 21081010241.-bab3.pdf Restricted to Repository staff only until 5 December 2028. Download (973kB)
	Text (Bab 4) 21081010241.-bab4.pdf Restricted to Repository staff only until 5 December 2028. Download (15MB)
	Text (Bab 5) 21081010241.-bab5.pdf Download (202kB)
	Text (Daftar pustaka) 21081010241.-daftarpustaka.pdf Download (147kB)
	Text (Lampiran) 21081010241.-lampiran.pdf Restricted to Repository staff only until 2028. Download (400kB)

Abstract

Rice leaf diseases such as Bacterial Leaf Blight, Brown Spot, Blast, and Tungro can significantly reduce crop productivity if not detected early. Manual identification by farmers is often inaccurate due to the visual similarity of symptoms among diseases, creating the need for an automated, image-based detection method. This study compares the performance of two Convolutional Neural Network architectures—ResNet50 and EfficientNetB0—in classifying four types of rice leaf diseases using the Rice Leaf Disease Images dataset from Kaggle. Five optimizers—Adam, Nadam, Adamax, SGD, and RMSprop—were evaluated for each architecture with learning rates of 0.001 and 0.0001 and training durations of 20 and 30 epochs. The training process employed transfer learning, data augmentation, and dataset splitting into 70% training, 20% validation, and 10% testing. The evaluation results show that the best performance was achieved by the ResNet50 model combined with the Adamax optimizer, a learning rate of 0.001, and 30 epochs, yielding a test accuracy of 97% with stable metric performance. This best-performing model was then implemented into an Android application (SiPadi), enabling users to upload rice leaf images and obtain automatic disease predictions. The findings demonstrate that selecting an appropriate CNN architecture and optimizer greatly influences model accuracy and stability in rice leaf disease classification.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	WAHANANI, HENNI ENDAH	NIDN0022097811	henniendah.if@upnjatim.ac.id
Thesis advisor	ANGGRAENY, FETTY TRI	NIDN0711028201	fettyanggraeny.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Muchammad Syamsu Huda

Date Deposited:

05 Dec 2025 09:36

Last Modified:

08 Dec 2025 02:34

URI:

https://repository.upnjatim.ac.id/id/eprint/48098