Egg Freshness Prediction Model Using Real-Time Cold Chain Storage Condition Based on Transfer Learning

Area of Science:

• Food Science
• Artificial Intelligence
• Agricultural Technology

Background:

• Egg quality monitoring is crucial during cold chain logistics.
• Variations in temperature and humidity affect egg quality.
• Haugh unit (HU) is a key indicator of egg freshness.

Purpose of the Study:

• To develop a deep learning model for predicting egg Haugh units (HU).
• To utilize non-destructive weight loss measurements for freshness prediction.
• To assess the model's performance in real-time cold chain environments.

Main Methods:

• A deep learning model combining CNN and LSTM was developed.
• Transfer learning was applied using temperature and weight loss data.
• Data augmentation techniques were used to enhance the dataset.
• Hyperparameter optimization was performed for the CNN-LSTM model.
• Performance was compared against general machine learning algorithms.

Main Results:

• The transfer learning CNN-LSTM model significantly reduced RMSE from 6.62 to 2.02 for HU prediction using weight loss data.
• Mean Absolute Error (MAE) decreased from 3.16 to 1.39 with data augmentation.
• The deep learning approach outperformed traditional machine learning methods.

Conclusions:

• Deep learning, specifically CNN-LSTM with transfer learning, offers a robust method for predicting egg freshness (HU).
• Non-destructive weight loss is a viable parameter for real-time egg quality monitoring in cold chains.
• The developed model can be implemented for practical applications in the egg industry.