Rancang Bangun Klasifikasi Kematangan Buah Durian Secara Multimodal Menggunakan Citra dan Aroma dengan Metode Convolutional Neural Network

Rahmadani, Muhammad Faiz (2024) Rancang Bangun Klasifikasi Kematangan Buah Durian Secara Multimodal Menggunakan Citra dan Aroma dengan Metode Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5009201112-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2027. Download (7MB) | Request a copy

Abstract

Buah Durian adalah salah satu buah tropis yang begitu populer di Indonesia dan telah dibudidayakan secara luas. Kualitas dari buah durian ditentukan oleh tingkat kematangannya, yang umumnya menggunakan metode konvensional, seperti membuka buah. Penelitian mengenai kematangan buah umumnya ditinjau dari warna kulit dan aroma dari buah, karena terdapat korelasi keduanya. Penelitian ini menggunakan metode convolutional neural network dengan data dari buah durian berupa gambar dan tegangan sensor gas. Tujuan penelitian ini adalah merancang sistem klasifikasi kematangan buah durian berdasarkan data gambar dan aroma secara multimodal dan menganalisis sistem klasifikasi tersebut berdasarkan akurasi prediksi. Perangkat keras yang digunakan terdiri dari Raspberry Pi, Arduino Nano, Kamera Logitech C270, Sensor MQ-3, dan ADS1115. Perangkat lunak secara umum dirancang menggunakan bahasa pemrograman Python dan Javascript yang terdiri dari algoritma preprocessing, model CNN citra dan aroma, dan Human Machine Interface. Pengambilan data buah durian dilakukan di daerah Trawas Kabupaten Mojokerto dengan buah durian jenis merica. Akurasi model CNN citra sebesar 87% dan model CNN aroma sebesar 100%, kurangnya akurasi model CNN citra disebabkan kurangnya dataset yang digunakan. Multimodal disempurnakan dengan weighted average untuk menggabungkan hasil prediksi dari enam model citra dan aroma serta didapatkan akurasi sebesar 100%.
====================================================================================================================================
Durian fruit is a tropical fruit that is very popular in Indonesia and has been widely cultivated. The quality of durian fruit is determined by the level of ripeness, which generally uses conventional methods, such as opening the fruit. Research on fruit ripeness is generally viewed from the skin color and aroma of the fruit, because there is a correlation between the two. This research uses a convolutional neural network method with data from durian fruit in the form of images and gas sensor voltage. The aim of this research is to design a classification system for durian fruit maturity based on multimodal image and aroma data and analyze the classification system based on prediction accuracy. The hardware used consists of Raspberry Pi, Arduino Nano, Logitech C270 Camera, MQ-3 Sensor, and ADS1115. The software is generally designed using the Python and Javascript programming languages which consist of preprocessing algorithms, CNN image and scent models, and a Human Machine Interface. Data collection on durian fruit was carried out in the Trawas area, Mojokerto Regency with merica durian fruit. The accuracy of the image CNN model is 87% and the aroma CNN model is 100%. The lack of accuracy of the image CNN model is due to the lack of datasets used. Multimodal was refined with a weighted average to combine the prediction results from six image and aroma models and obtained an accuracy of 100%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Accuracy, Convolutional Neural Network, Durian, Gas Sensor, Ripeness, Akurasi, Convolutional Neural Network, Durian, Kematangan, Sensor Gas
Subjects:	T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Faiz Rahmadani
Date Deposited:	22 Jan 2025 03:31
Last Modified:	22 Jan 2025 03:31
URI:	http://repository.its.ac.id/id/eprint/116576

Actions (login required)

View Item