Desenvolvimento e validação de um software para avaliação da qualidade da carne em tempo real utilizando o VIS-NIRS

Abstract

The validation of auxiliary tools to evaluate meat quality in real time in the industry will bring great advances to the production process with the possibility of increasing the percentage of cuts classified by quality and adding value to the product. The aim of the study was to develop and validate a software to predict beef quality in real time, based on information generated by VIS-NIRS. The Python 3.0 programming language and the Qt Designer 5.0 tool were used in the software development to design the graphical interface. Data on meat characteristics of 430 Nellore and crossbred cattle randomly evaluated in a commercial slaughterhouse were used in the validation. Meat characteristics were evaluated in Longissimus thoracis between the 5th and 6th ribs to determine attributes related to quality such as pH24, tenderness with maturation of two and seven days, marbling and ethereal extract. Near-visible wavelength infrared assessments (VIS-NIRS) were taken for association with reference variables. SoftBeef 1.3 was developed for desktop, has user-friendly interfaces, and can be installed on different operating systems (interoperable). SoftBeef is able to perform transmittance file import, convert transmittance files to absorbance, preprocess data, analyze and predict in real time. The models generated for pH24 and tenderness after two days of maturation (R²p -0.319 and R²p -0.078, respectively) were not satisfactory for predicting the characteristics. SoftBeef 1.3 presented limitations during the analysis and is in the adjustment phase.