Stendafity, Selfi (2024) Pengembangan Sistem Pencitra untuk Identifikasi dan Estimasi Adulteran pada Daging Sapi Cincang Menggunakan Deep Neural Network. Masters thesis, Institut Teknologi Sepuluh Nopember.
Text
6009221009-MASTER_THESIS.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (11MB) | Request a copy |
Abstract
Daging cincang adalah bahan baku penting dari banyak bahan makanan yang dikonsumsi secara luas di seluruh dunia. Karena tingginya permintaan dan harga yang mahal, mengakibatkan daging cincang rentan terhadap pemalsuan. Karena daging cincang akan kehilangan karakteristik morfologi dari karakteristik daging mentah biasa, sehingga akan sulit untuk mengidentifikasi daging cincang yang sudah dicampur berdasarkan warna dan teksturnya. Penelitian ini mengembangkan sistem pencitra untuk mengidentifikasi dan estimasi kandungan adulterant pada daging sapi cincang dengan menggunakan jaringan syaraf tiruan. Data yang digunakan adalah data citra, yang kemudian, dilakukan ekstraksi fitur warna dan tekstur sebagai input dari model yang akan dikembangkan. Dalam penelitian ini, terdapat tiga tahapan model, yaitu pertama membuat model klasifikasi daging sapi cincang murni dan yang sudah dipalsukan. Kedua adalah mengembangkan model untuk klasifikasi jenis adulterant yang terkandung dalam daging sapi cincang dan yang ketiga adalah membuat model untuk estimasi prosentase kandungan adulterant pada daging sapi cincang. Pada model pertama diperoleh akurasi sebesar 99.85%, presisi dengan rentang 97,87-100%, dan sensitivitas 96,84-100%. Kemudian pada model kedua, diperoleh hasil akurasi sebesar 99,64%, presisi dengan rentang 98,96-100%, dan sensitivitas sebesar 99,42-100%. Sedangkan model ketiga diperoleh nilai R2 dengan range 0,9668-0,9957, MAE 0,0061-0,0094, MSE sebesar 0.00011-0.00082, dan RMSE sebesar 0,010-0,029.
======================================================================================================
Minced meat is an important raw material of many food items that are widely consumed around the world. Due to its high demand and high price, minced meat is prone to adulteration. Because minced meat will lose the morphological characteristics of ordinary raw meat characteristics, it will be difficult to identify minced meat that has been mixed based on its color and texture. This research develops an imaging system to identify and estimate the adulterant content in minced beef using artificial neural networks. The data used is image data, which is then extracted from color and texture features as input for the model to be developed. In this research, there are three stages of the model, the first is to create a classification model for pure and adulterated minced beef. The second is to develop a model for the classification of adulterant types contained in minced beef and the third is to create a model for estimating the percentage of adulterant content in minced beef. The first model obtained an accuracy of 99.85%, precision with a range of 97.87-100%, and sensitivity of 96.84-100%. Then in the second model, the accuracy of 99.64%, precision with a range of 98.96-100%, and sensitivity of 99.42-100% were obtained. While the third model obtained R2 value with a range of 0.9668-0.9957, MAE 0.0061-0.0094, MSE of 0.00011-0.00082, and RMSE of 0.010-0.029.
Item Type: | Thesis (Masters) |
---|---|
Uncontrolled Keywords: | Adulterant, minced beef, neural network, imaging system, Adulteran, daging sapi cincang, neural network, sistem pencitra |
Subjects: | Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines. Q Science > Q Science (General) > Q337.5 Pattern recognition systems T Technology > T Technology (General) > T57.5 Data Processing T Technology > TR Photography |
Divisions: | Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30101-(S2) Master Thesis |
Depositing User: | Selfi Stendafity |
Date Deposited: | 11 Aug 2024 12:28 |
Last Modified: | 11 Aug 2024 12:28 |
URI: | http://repository.its.ac.id/id/eprint/114557 |
Actions (login required)
View Item |