Doni, Raviqi Ikhsan (2026) Implementasi metode jaringan saraf tiruan untuk klasifikasi kekurangan nutrisi tanaman berdasarkan citra daun. Sarjana thesis, UIN Sunan Gunung Djati Bandung.
|
Text
1_cover.pdf Download (135kB) | Preview |
|
|
Text
2_abstrak.pdf Download (221kB) | Preview |
|
|
Text
3_skbebasplagiarism.pdf Download (391kB) | Preview |
|
|
Text
4_daftarisi.pdf Download (190kB) | Preview |
|
|
Text
5_bab1.pdf Download (513kB) | Preview |
|
![[img]](https://digilib.uinsgd.ac.id/style/images/fileicons/text.png) |
Text
6_bab2.pdf Restricted to Registered users only Download (454kB) |
|
|
Text
7_bab3.pdf Restricted to Registered users only Download (322kB) |
|
|
Text
8_bab4.pdf Restricted to Registered users only Download (558kB) |
|
|
Text
9_bab5.pdf Restricted to Registered users only Download (449kB) |
|
|
Text
10_bab6.pdf Restricted to Registered users only Download (205kB) |
|
|
Text
11_daftarpustaka.pdf Restricted to Registered users only Download (196kB) |
|
|
Text
12_lampiran.pdf Restricted to Repository staff only Download (395kB) |
Abstract
INDONESIA : Kekurangan nutrisi pada tanaman padi merupakan salah satu faktor yang dapat menurunkan produktivitas dan kualitas hasil panen. Identifikasi defisiensi nutrisi secara manual masih mengandalkan observasi visual yang membutuhkan keahlian khusus dan sering kali tidak akurat. Penelitian ini bertujuan untuk mengembangkan sistem klasifikasi kekurangan nutrisi tanaman padi berbasis citra daun menggunakan metode Jaringan Saraf Tiruan dengan pendekatan transfer learning pada arsitektur EfficientNet-B0. Dataset yang digunakan berasal dari Kaggle dengan total 1.156 citra yang terdiri dari tiga kelas defisiensi nutrisi, yaitu Nitrogen (N), Fosfor (P), dan Kalium (K). Tahapan penelitian meliputi preprocessing (resize, normalisasi, dan augmentasi), pengembangan model dengan menambahkan lapisan GlobalAveragePooling2D, Dropout, dan Dense Softmax, serta evaluasi menggunakan metrik akurasi, presisi, recall, dan F1-score. Hasil penelitian menunjukkan bahwa model EfficientNet-B0 mampu melakukan klasifikasi dengan kinerja yang sangat baik, ditandai dengan nilai validation accuracy sebesar 97% dan validation loss sekitar 0.2. Evaluasi confusion matrix serta metrik klasifikasi menghasilkan nilai ratarata precision, recall, dan F1-score sebesar 0.95, yang menandakan konsistensi kinerja pada seluruh kelas. Model juga mampu memberikan prediksi citra baru secara akurat disertai confidence score. Dengan demikian, penelitian ini menunjukkan bahwa metode jaringan saraf tiruan berbasis EfficientNet-B0 berpotensi diterapkan sebagai alat bantu klasifikasi dini kekurangan nutrisi tanaman padi yang cepat, akurat, dan efisien. ENGLISH: Nutrient deficiency in rice plants is one of the factors that can reduce productivity and crop quality. Manual identification of nutrient deficiencies still relies on visual observation, which requires specific expertise and is often inaccurate. This study aims to develop a system for classifying nutrient deficiencies in rice plants based on leaf images using an Artificial Neural Network with a transfer learning approach on the EfficientNet-B0 architecture. The dataset used was sourced from Kaggle, comprising a total of 1,156 images divided into three nutrient deficiency classes: Nitrogen (N), Phosphorus (P), and Potassium (K). The research stages included preprocessing (resizing, normalization, and augmentation), model development by adding GlobalAveragePooling2D, Dropout, and Dense Softmax layers, and evaluation using accuracy, precision, recall, and F1-score metrics. The results show that the EfficientNet-B0 model achieved excellent classification performance, indicated by a validation accuracy of 97% and a validation loss of approximately 0.2. Evaluation using the confusion matrix and classification metrics yielded an average precision, recall, and F1-score of 0.95, indicating consistent performance across all classes. The model was also able to accurately predict new images along with a confidence score. Thus, this study demonstrates that the EfficientNet-B0-based artificial neural network method has the potential to be applied as a rapid, accurate, and efficient tool for the early detection of nutrient deficiencies in rice plants.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Uncontrolled Keywords: | Artificial Neural Network; EfficientNet-B0; Transfer Learning; Leaf Image Nutrient Deficiency; Rice Plants |
| Subjects: | Special Computer Methods Special Computer Methods > Artificial Intelligence Applied Physics Applied Physics > Electrical Engineering |
| Divisions: | Fakultas Sains dan Teknologi > Program Studi Teknik Elektro |
| Depositing User: | Raviqi Ikhsan Doni |
| Date Deposited: | 05 Oct 2026 07:31 |
| Last Modified: | 07 Oct 2026 02:40 |
| URI: | https://digilib.uinsgd.ac.id/id/eprint/142933 |
Actions (login required)
 |
View Item |
