Quality Monitoring System For Pork Meat Using Computer Vision

To realize the purpose of the study, the researchers implemented and evaluated a system that estimates the age of pork meat through the RGB features of its various segments whose images were captured using a custom-built platform. Furthermore, the study aimed to establish a relationship between the colors of a pork sample's fat, meat, and miscellaneous area to estimate the age of the sample using computer vision techniques. In order to produce a reliable data model, the software's dataset was established using sufficient number of pork meat images comprising specimens sampled on different days to include ages ranging from 0-7 days. Pork meat samples from each age group were individually placed in a specially-designed platform equipped with proper lighting and a high-resolution camera for taking photographs. Images of the samples were captured and subjected to image processing in order to produce data through feature extraction. By automating the image analysis of the entire data set, extracted data were processed and analyzed in order to reliably estimate the age of pork meat samples in a non-destructive manner. As part of the evaluation, this study determined the accuracy of three regression models: Linear, Multiple, and Polynomial regressions in estimating the age of pork meat. Upon evaluation, the result established that the age of pork meat can be estimated by the color of its fat. However, estimates were significantly improved by also using its meat and miscellaneous area. It could be surmised that using only the color of the fat is sufficient but would make estimations less accurate as compared to using all the aforementioned parts. Therefore, it is concluded that using the color of fat, combined with that of meat and other miscellaneous parts, the accuracy of estimating the age of pork meat can be improved. The system's quality was also evaluated according to the standards for computer software set by ISO 25010 International Quality Standards. The over-all mean of 4.96 establishes that the system has "excellent" quality in terms of functional suitability, reliability, performance efficiency, usability, security, compatibility, maintainability, and portability.