Automatic ham classification method based on support vector machine model increases accuracy and benefits compared to...

Area of Science:

• Animal Science
• Food Science
• Agricultural Engineering

Background:

• Subcutaneous fat thickness (SFT) is crucial for determining ham processing methods.
• Accurate SFT prediction is essential for efficient slaughter line operations.
• Current manual SFT measurement methods may lack precision and economic efficiency.

Purpose of the Study:

• To compare the accuracy and economic benefits of an automatic Support Vector Machine (SVM) model against manual measurement for predicting SFT in hams.
• To evaluate the performance of an SVM-based system for SFT classification on the slaughter line.
• To determine the potential economic advantages of implementing an SVM model for ham sorting.

Main Methods:

• A Support Vector Machine (SVM) model was developed for automatic SFT prediction.
• Manual SFT measurements were performed by an experienced operator.
• Both methods were validated against a gold standard (ruler measurement) using 400 hams across different SFT classes.

Main Results:

• The SVM model achieved a prediction accuracy of 75.3% for SFT.
• This represents a 5.5% accuracy improvement over manual measurement.
• The SVM model is projected to increase economic benefits by 12-17% for ham sorting.

Conclusions:

• SVM classification offers superior accuracy for predicting subcutaneous fat thickness compared to manual methods.
• Implementing SVM models in slaughter lines enhances economic benefits through improved ham sorting.
• Automatic SFT prediction using SVM provides a more precise and profitable approach to ham processing.