Detection of adulteration in mutton using digital images in time domain combined with deep learning algorithm

Area of Science:

• Food Science and Technology
• Artificial Intelligence in Food Analysis
• Machine Learning for Quality Control

Background:

• Mutton adulteration poses a significant challenge to food authenticity and consumer safety.
• Traditional methods for detecting adulteration can be time-consuming, labor-intensive, and may lack accuracy.
• Developing rapid, accurate, and non-destructive analytical techniques is crucial for the food industry.

Purpose of the Study:

• To propose and validate a novel method for the qualitative discrimination and quantitative prediction of adulterated mutton.
• To investigate the effectiveness of using digital images in the time domain combined with convolutional neural networks (CNNs).
• To assess the impact of temperature variations during the heating process on detection performance.

Main Methods:

• Acquisition of 195 sample images during a 10-minute constant temperature heating process.
• Application of a convolutional neural network (CNN) model for qualitative discrimination and quantitative prediction.
• Comparison of model performance between the initial heating stage and the entire heating process to evaluate temperature disturbance effects.

Main Results:

• The CNN model utilizing the entire heating process data demonstrated superior performance compared to using only the initial heating stage.
• Qualitative discriminant model accuracy increased by 7.33% when analyzing the full heating process.
• Quantitative prediction models for duck/pork adulteration showed improvements: R² increased by 0.08/0.07, RPD by 0.85/0.87, and RMSE decreased by 0.01/0.01.

Conclusions:

• The proposed method, integrating digital imaging and CNNs, effectively detects and analyzes adulterated mutton.
• Utilizing temperature fluctuations during the heating process enhances the detection capabilities of the analytical model.
• This approach offers a promising, accurate, and efficient solution for ensuring mutton authenticity and quality control.