Modeling of Ethiopian Beef Meat Marbling Score Using Image Processing for Rapid Meat Grading
Tariku Erena1, Abera Belay1, Demelash Hailu1
1Department of Food Science and Applied Nutrition, Bioprocessing and Biotechnology Center of Excellence, Addis Ababa Science and Technology University, Addis Ababa P.O. Box 16417, Ethiopia.
Journal of Imaging
|June 26, 2024
Summary
This study developed a digital tool using image processing and machine learning to predict beef marbling scores in Ethiopian cattle breeds. The findings confirm the link between marbling and meat quality attributes.
Area of Science:
- Animal Science
- Food Science
- Computer Science
Background:
- High marbling in beef is linked to superior sensory qualities.
- Predicting marbling accurately is crucial for the meat industry.
- Ethiopian indigenous cattle breeds like Boran, Senga, and Sheko require quality assessment.
Purpose of the Study:
- To predict marbling scores in Boran, Senga, and Sheko cattle using digital image processing.
- To analyze the relationship between marbling, texture, and sensory attributes.
- To develop a digital tool for objective beef quality evaluation.
Main Methods:
- Digital image processing was used to analyze marbling in Longissimus dorsi muscle.
- An extreme gradient boosting (GBoost) machine learning algorithm predicted marbling scores.
- Meat texture and sensory characteristics were evaluated using a texture analyzer and trained sensory panel.
Main Results:
- The GBoost model accurately predicted marbling scores with R² = 0.83.
- Boran cattle exhibited the highest fat content and marbling scores.
- Significant differences in tenderness and sensory attributes were observed among the breeds.
Conclusions:
- A digital tool for predicting beef marbling scores in Ethiopian cattle breeds was successfully developed.
- The study validated the correlation between beef marbling and key quality attributes.
- The developed technology holds potential for application in the meat industry and quality control.
