Imaging System-Based Pork Adulteration Detection in Minced Beef using Neural Network

• Main Authors: Stendafity, Selfi, Suyanto, Hatta, Agus M., Setiadi, Iwan C., Pertiwi, Nabilah Kartika, Anisa Damayanti, Ayu, Hartati, Ayu Dian, Dwi Susanti, Mia, Shoffiana, Nur Alfiani
• Format: Conference Proceeding
• Language: English
• Subjects: Artificial neural networks; Computational modeling; Food security; Imaging; imaging system; Mechatronics; minced beef; neural network; pork adulteration; Regulation; Root mean square; Sensitivity
• ISSN: 2832-8353
• DOI: 10.1109/ICAMIMIA60881.2023.10427668

The high demand and related high cost of minced meat make it vulnerable to adulteration. The practice of adulteration can have detrimental implications for individuals' well-being and can provide significant challenges for religious communities adhering to dietary regulations. For example, individuals who follow the Islamic faith are forbidden from consuming pork, hence emphasizing the significance of swiftly determining the existence of pork in the meat they ingest. This study aimed to examine the use of an imaging technology for the detection and quantification of pork adulteration in raw minced beef. The initial approach employed for the categorization of unadulterated beef, beef contaminated with pork, and unadulterated pork is artificial neural networks, which attains a remarkable accuracy rate of 99.01%. The second model employed deep neural networks to evaluate the level of adulteration in minced beef with pork. This model achieved a root mean square error (RMSE) value of 0.01681 and a coefficient of determination (R 2 ) value of 0.98890.