COMPARATIVE ANALYSIS OF PHOTO CLASSIFICATION MODELS FOR RICE GROWTH PHASE USING CNN, RESNET AND VGG ALGORITHM (CASE STUDY: KSA SURVEY OF STATISTICS INDONESIA)

• Main Author: Tri Hutami, Ratih
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/84846
• Institution: Institut Teknologi Bandung

The classification of rice growth phases is needed to calculate the statistics for the harvested area of rice and rice production. Accurate rice production statistics are needed in making government decisions regarding efforts to realize people's food sovereignty. In the integrated food crop agricultural statistical data with the area sample frame method (KSA survey), classification is done manually by the officer and manual validation by the examiner. As a result, errors in filling the rice growth phase are still found, which can lead to errors in estimating rice production. In an era in which machine learning is highly developed, the image classification process can be automated. The algorithm that is considered sufficiently capable for image classification is the Convolutional Neural Network (CNN) algorithm. In this study, a classification model will also be tested using the ResNet and VGG architectures, which are developments of CNN. The best model to be used for the data validation process. The research method used is based on the Design Research Methodology.