MotivNet: Emotionally Intelligent Foundation Model for Facial Emotion Recognition

This paper introduces MotivNet, a facial emotion recognition (FER) model designed for real-world application. It addresses the generalization problem of existing FER models by leveraging the Meta-Sapiens foundation model, which is pre-trained on a large scale. The key contribution is achieving competitive performance across diverse datasets without cross-domain training, a common limitation of other approaches. This makes FER more practical for real-world use.

• •MotivNet is a facial emotion recognition model designed for real-world application.
• •It leverages the Meta-Sapiens foundation model for improved generalization.
• •Achieves competitive performance without cross-domain training.
• •The code is publicly available.