Analysis of facial emotion expression in eating occasions using deep learning

Area of Science:

• Psychology
• Computer Science
• Neuroscience

Background:

• Eating is an emotional and social activity influenced by various factors.
• Emotions during eating impact health conditions like obesity.
• Assessing eating-related emotions is a multidisciplinary challenge.

Purpose of the Study:

• To automatically evaluate emotional experiences during eating using deep learning.
• To develop a facial expression-based model to reduce user bias and errors in emotion assessment.
• To investigate the influence of external factors (videos, music) on emotions during eating.

Main Methods:

• Utilized a mobile app to capture neural, behavioral, and physical emotional manifestations.
• Employed facial expression recognition through three Convolutional Neural Network (CNN) architectures: fine-tuned VGG16 and Deepface.
• Tested emotional responses under three conditions: watching videos, listening to music, or doing nothing while eating.

Main Results:

• External factors, not just food, were found to significantly affect emotional changes during eating.
• The fine-tuned VGG16 model achieved 77.68% accuracy in recognizing four distinct emotions.
• The developed system offers a potential alternative to subjective survey-based food evaluation methods.

Conclusions:

• Facial expression analysis provides an objective method for assessing emotions during eating.
• Environmental stimuli play a crucial role in modulating the emotional experience of eating.
• Deep learning models show promise for automated, unbiased evaluation of food-related emotions.