METODE CONVOLUTIONAL NEURAL NETWORK SEBAGAI IDENTIFIKASI KEBUTUHAN NUTRISI TANAMAN PADI HIDROPONIK
Total View This Week0
Total View This Week0
Institusion
Institut Teknologi Perusahaan Listrik Negara
Author
Safiera, Nurin Masyitha
Siregar, Riki Ruli A.
Haris, Abdul
Siregar, Riki Ruli A.
Haris, Abdul
Subject
Teknik Informatika
Datestamp
2023-06-20 06:54:35
Abstract :
Rice is one of the most important staple crops in the world. Most regions in the world make rice as a staple food, especially in Indonesia. Nutrition is one of the important things in plant growth and development. Lack of nutrients in plants can affect the growth process and the quality of plants that are ready to be harvested. The Convolutional Neural Network method was chosen to identify the nutritional needs of hydroponic rice plants. In this study, the dataset used was 1190. Rice leaf images which were divided into 2 classes, Enough and Less, which were tested with a comparison of 80% as data train, 10% as data test, and 10% as validation data. There are three architecture models used in this research, VGG16, MobileNet, and Xception using Jupyter and Google Colaboratory as tools. The training process was carried out by entering 10 epochs and batch sizes of 32 and 64. The best accuracy results obtained were VGG16 78,15% and 86,47%, MobileNet of 82,69% and 86,55%, Xception of 82,33% and 88,24%. Meanwhile, the best overall accuracy results were obtained from the Xception model of 88,24% with batch size onput of 32, and the tools used were Jupyter.
Rice is one of the most important staple crops in the world. Most regions in the world make rice as a staple food, especially in Indonesia. Nutrition is one of the important things in plant growth and development. Lack of nutrients in plants can affect the growth process and the quality of plants that are ready to be harvested. The Convolutional Neural Network method was chosen to identify the nutritional needs of hydroponic rice plants. In this study, the dataset used was 1190. Rice leaf images which were divided into 2 classes, Enough and Less, which were tested with a comparison of 80% as data train, 10% as data test, and 10% as validation data. There are three architecture models used in this research, VGG16, MobileNet, and Xception using Jupyter and Google Colaboratory as tools. The training process was carried out by entering 10 epochs and batch sizes of 32 and 64. The best accuracy results obtained were VGG16 78,15% and 86,47%, MobileNet of 82,69% and 86,55%, Xception of 82,33% and 88,24%. Meanwhile, the best overall accuracy results were obtained from the Xception model of 88,24% with batch size onput of 32, and the tools used were Jupyter.
Download
book
BibTex
Latex, Jabref
cloud_download
Original Resource
url resource Institution
