Azhar, Moch Fadillah (2026) Implementasi algoritma Convolutional Neural Network dengan arsitektur ResNet-50 pada klasifikasi penyakit daun stroberi. Sarjana thesis, UIN Sunan Gunung Djati Bandung.
Full text not available from this repository. (Request a copy)Abstract
ABSTRAK Penyakit pada daun stroberi merupakan salah satu faktor utama yang dapat menurunkan kualitas dan produktivitas tanaman stroberi. Identifikasi penyakit daun secara manual memerlukan keahlian khusus serta berpotensi menimbulkan kesalahan akibat keterbatasan pengamatan visual. Oleh karena itu, penelitian ini bertujuan untuk mengembangkan sistem klasifikasi penyakit daun stroberi berbasis citra digital menggunakan metode Deep Learning dengan arsitektur Convolutional Neural Network (CNN) ResNet50. Dataset yang digunakan terdiri dari lima kelas, yaitu Angular Leafspot, Healty, Leaf Blight, Leaf Spot, dan Powdery Mildew Leaf. Penelitian ini menerapkan pendekatan transfer learning dengan strategi pelatihan dua tahap, yaitu training awal (base frozen) dan fine-tuning. Beberapa skenario pembagian dataset diuji, meliputi rasio 60:20:20, 70:15:15, dan 80:10:10, untuk mengetahui pengaruh rasio data terhadap kinerja model. Evaluasi model dilakukan menggunakan metrik accurasy, loss, precision, recall, dan f1-score. Hasil penelitian menunjukkan bahwa model ResNet50 mampu melakukan klasifikasi penyakit daun stroberi dengan performa yang sangat baik. Skenario 80:10:10 dengan fine-tuning menghasilkan kinerja terbaik dengan akurasi sebesar 98,40%, serta nilai precision, recall, dan f1-score masing-masing sebesar 0,98. Berdasarkan hasil tersebut, dapat disimpulkan bahwa metode CNN berbasis ResNet50 efektif digunakan untuk klasifikasi penyakit daun stroberi dan berpotensi dikembangkan sebagai sistem pendukung deteksi dini penyakit tanaman berbasis citra digital. ABSTRACT Diseases affecting strawberry leaves are one of the main factors that reduce the quality and productivity of strawberry plants. Manual identification of leaf diseases requires specific expertise and is prone to errors due to limitations in visual observation. Therefore, this study aims to develop an image-based strawberry leaf disease classification system using Deep Learning with a Convolutional Neural Network (CNN) ResNet50 architecture. The dataset used in this study consists of five classes, namely Angular Leafspot, Healty, Leaf Blight, Leaf Spot, and Powdery Mildew Leaf. This research applies a transfer learning approach with a two-stage training strategy, namely initial training (base frozen) and fine-tuning. Several dataset split scenarios were evaluated, including 60:20:20, 70:15:15, and 80:10:10, to analyze the effect of data ratio on model performance. Model evaluation was conducted using accuracy, loss, precision, recall, and f1-score metrics. The experimental results demonstrate that the ResNet50 model is capable of classifying strawberry leaf diseases with excellent performance. The 80:10:10 dataset split with fine-tuning achieved the best results, obtaining an accuracy of 98.40%, with precision, recall, and f1-score values of 0.98. Based on these results, it can be concluded that the CNN-based ResNet50 method is effective for strawberry leaf disease classification and has strong potential to be developed as a decision support system for early plant disease detection using digital images.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Subjects: | Data Processing, Computer Science > Systems Analysis and Computer Design |
| Divisions: | Fakultas Sains dan Teknologi > Program Studi Teknik Informatika |
| Depositing User: | Moch Fadillah Azhar |
| Date Deposited: | 16 Mar 2026 02:02 |
| Last Modified: | 16 Mar 2026 02:02 |
| URI: | https://digilib.uinsgd.ac.id/id/eprint/128955 |
Actions (login required)
 |
View Item |
