KLASIFIKASI PENYAKIT DAUN PADI DENGAN MENGGUNAKAN PRETRAINED MODEL CNN RESNET 34

RAMDHANI, ALIF FIRMAN (2023) KLASIFIKASI PENYAKIT DAUN PADI DENGAN MENGGUNAKAN PRETRAINED MODEL CNN RESNET 34. S1 thesis, Universitas Mercu Buana Bekasi.

Preview	Text 41519210043 - Alif Firman Ramdhani - 01 Cover.pdf Download (341kB) | Preview
Preview	Text 41519210043 - Alif Firman Ramdhani - 02 Abstrak.pdf Download (136kB) | Preview
	Text 41519210043 - Alif Firman Ramdhani - 03 BAB 1.pdf Restricted to Registered users only Download (167kB)
	Text 41519210043 - Alif Firman Ramdhani - 04 BAB 2.pdf Restricted to Registered users only Download (330kB)
	Text 41519210043 - Alif Firman Ramdhani - 05 BAB 3.pdf Restricted to Registered users only Download (208kB)
	Text 41519210043 - Alif Firman Ramdhani - 06 BAB 4.pdf Restricted to Registered users only Download (608kB)
	Text 41519210043 - Alif Firman Ramdhani - 07 BAB 5.pdf Restricted to Registered users only Download (116kB)
	Text 41519210043 - Alif Firman Ramdhani - 08 Daftar Pustaka.pdf Restricted to Registered users only Download (169kB)
	Text 41519210043 - Alif Firman Ramdhani - 09 Lampiran.pdf Restricted to Registered users only Download (1MB)

Abstract

Penyakit merupakan salah satu penyebab kegagalan panen padi. Deteksi penyakit tanaman antara lain dapat dilakukan dengan Teknik data science. Tujuan dari penelitian ini adalah untuk menentukan empat jenis penyakit tanaman padi yaitu brown spot, bacterial leaf blight, leaf blast, dan sheat blight menggunakan algoritma pretrained model CNN ResNet34. Dataset berupa 1396 gambar yang diperoleh dari data publik yang terdapat pada www.kaggle.com. Tahapan preprocessing terdiri dari resize, normalize, dan convert to tensor. Tahap pemodelan terdiri dari lapisan konvolusi, lapisan normalisasi batch, dan lapisan non-linear seperti ReLU (Rectified Linear Unit). Parameter uji yang digunakan berupa rasio pembagian data, optimizer, batch size, dan crop size untuk mengukur pengaruh pada sistem berupa nilai akurasi, precision, recall, f1 score. Dilakukan pembangunan model dengan pembagian data testing adalah 60:40, 70:30, 80:20 dan 90:10, optimizer berupa Adam, AdamW, RMSprop, batch size dengan ukuran 16, 32, 64 dan crop size sebesar 221x221, 224x224, 227x227. Pada penelitian ini didapatkan hasil terbaik dengan menggunakan rasio 80:20, optimizer RMSprop, batch size 32, crop size 224x224 dengan akurasi 99,69%, precision, recall, f1 score sebesar 100%. Kata kunci: tanaman padi,deteksi penyakit,convolutional neural network,resnet34 Diseases are one of the causes of paddy crop failure. Plant disease detection can be performed using data science techniques. The aim of this research is to identify four types of paddy plant diseases, namely brown spot, bacterial leaf blight, leaf blast, and sheath blight, using the pretrained CNN ResNet34 model algorithm. The dataset consists of 1396 images obtained from public data available on www.kaggle.com. The preprocessing stage involves resizing, normalization, and converting to tensors. The modeling stage consists of convolutional layers, batch normalization layers, and non-linear layers such as Rectified Linear Unit (ReLU). The evaluation parameters used include data split ratio, optimizer, batch size, and crop size to measure the impact on the system in terms of accuracy, precision, recall, and F1 score. The model is built with data splits of 60:40, 70:30, 80:20, and 90:10, using optimizers such as Adam, AdamW, RMSprop, and batch sizes of 16, 32, 64, and crop sizes of 221x221, 224x224, 227x227. The research yielded the best results with a data split ratio of 80:20, RMSprop optimizer, batch size of 32, and crop size of 224x224, achieving an accuracy of 99,69% and precision, recall, and F1 score of 100%. Keywords: rice plant,disease detection,confolutional neural network,resnet34

Actions (login required)

View Item