WUR Babies: Emotion and Eating Behavior Analysis for Food Preference Prediction in Children (MSc)

Short description

This project is designed to develop a novel method for predicting children's food preferences, aiming to promote healthier eating habits. The approach involves refining an existing emotion detection model to accurately interpret children's emotional reactions associated with food. A key part of this process is the annotation of 50 videos, each 10 minutes long, showing children in various emotional states and eating behaviors. This analysis will form the basis of a training set that captures the subtle interplay between emotions and food consumption.

The student will then integrate this enhanced emotion detection system with an eating detection model. This integration is crucial for extracting significant features that link children's emotional expressions with their eating actions. Using these combined insights, the student will develop a new machine learning model specifically designed to predict food preferences in children. This model will be trained to discern patterns that connect emotional responses and eating behaviors to food likings.

The final phase involves validating this model using a separate set of data obtained from children of HNH employees. This step is vital to assess the model's real-world effectiveness and reliability, ensuring it can accurately predict children's food preferences by analyzing emotional and behavioral indicators.

Objectives

Train a deep learning model that can predict food liking in children (2 to 8 years old) from videos.

Tasks

The work in this MSc thesis entails:

• Refine an existing emotion detection model to accurately interpret children's emotional reactions associated with food
• Integrate this enhanced emotion detection system with an eating detection model
• Validate this model using a separate set of data obtained from children of HNH employees

Literature:

Requirements:

• Courses: “HNH-37006 Data Science for Health Principles” and/or “HNH39003 Data Types for Signal Processing – or similar experience obtained from other courses.
• Required skills/knowledge: Intermediate Machine Learning: Familiarity with core machine learning concepts, especially classification and regression, and experience with a relevant libraries like scikit-learn.

Key words: machine learning, data analysis, Python programming, computer vision, research methodology, problem-solving, critical thinking

Contact person(s)
Sjoukje Osinga (sjoukje.osinga@wur.nl)