Quality Assessment of Beef Using Computer Vision Technology.

Area of Science:

• Food Science and Technology
• Agricultural Engineering
• Meat Science

Background:

• Traditional methods for assessing meat quality are often destructive and time-consuming.
• Computer vision (CV) technology presents a promising alternative for rapid, non-destructive quality evaluation of agricultural products, including meat.
• The objective evaluation of beef quality attributes is crucial for the meat industry.

Purpose of the Study:

• To evaluate the efficacy of computer vision (CV) technology in predicting various quality attributes of beef.
• To establish correlations between image analysis data and physicochemical, proximate, biochemical, and microbiological parameters of beef.

Main Methods:

• Images of *longissimus dorsi* muscle from beef were captured 24 hours post-mortem.
• Quality traits including color (L*, a*, b*), pH, drip loss, cooking loss, dry matter, moisture, protein, fat, ash, TBARS, POV, FFA, and microbial counts (TCC, TVC, TYMC) were measured using standard laboratory techniques.
• Image analysis was performed using Matlab software, with calibration and validation models fitted using Unscrambler X software.

Main Results:

• A significant correlation was observed between the 'a*' color value from image analysis and the actual 'a*' value (r=0.65) and moisture content (r=0.56).
• The highest calibration (r²c=0.73) and prediction (r²p=0.69) accuracy was achieved for lightness (L*) prediction.
• CV technology showed potential for predicting multiple beef quality traits.

Conclusions:

• Computer vision technology demonstrates significant potential as a valuable tool for predicting beef quality attributes.
• This non-destructive technique can be effectively applied in both laboratory settings and meat processing industries for quality assessment.
• Further research can optimize CV models for broader application in meat quality control.