|
![](images/pix01.gif) |
|
|
|
Appropriate and Optimal Classifier for Beef Quality Discrimination by A Low-Cost Optical Apparatus |
|
PP: 2665-2677 |
|
doi:10.18576/isl/121016
|
|
Author(s) |
|
Said E. Abdallah,
Wael M. Elmessery,
Waleed Z. Hassan,
Nora S. Al-Sattary,
Amr A. Abohany,
Ahmed Elashry,
|
|
Abstract |
|
In this paper, we present an optimal classifier for beef quality discrimination by a low-cost optical apparatus. Detecting beef spoilage in beef factories is a sophisticated process because beef spoilage is a mixture of physical and chemical changes. A low-cost Light-Dependent Resistor (LDR), and a light source were used to collect reflection spectra during the analysis of beef. The LabVIEW platform was programmed to acquire the obtained data from the microcontroller (Arduino) to predict beef quality. For the beef quality discrimination process, un-supervising machine learning called Principal Components Analysis (PCA) was used, and the score plot percentage was of the first (F1) and second (F2) dimensions of the most variation for forty samples were of 93.98% and 3.38% respectively. Supervised Machine Learning (SML) (Support Vector Machine (SVM) and Linear Discriminant Analysis (LDA)) were used also to compare with other models of un-supervised machine learning. Optimum classifier was achieved by the classification algorithm of SVM that can represent 95.75% of the whole data.
|
|
|
![](images/pix01.gif) |
|
|