Valence and Arousal prediction from food images using Deep Learning and classical Machine Learning models

This thesis explores the prediction of valence and arousal from food images using both classical machine learning models (SVR, Random Forest) and a deep learning-based MLP implemented in PyTorch. The models are evaluated through cross-validation using standard regression metrics (MAE, MSE, R²). Results highlight the strengths and limitations of each approach in affective image analysis.