Landmark Driven Facial Expression Recognition
A landmark-driven method on Facial Expression Recognition (FER)
Overview
The Landmark-Driven Facial Expression Recognition system is a method based on the SCN backbone intended for facial expression recognition (FER). This method does not use official FER module implementations and is designed for research purposes. The integration of facial landmark detection algorithms into the FER framework has shown improved accuracy in experimental results, eliminating the need for pretrained models.
Features
- Landmark-driven method for Facial Expression Recognition
- Utilizes robust and efficient facial landmark detection algorithms
- Combines facial landmark detection and FER tasks in one framework
- Experimental results show the improvement of FER accuracy with facial landmarks
- Data from FER2013 dataset with image properties of 48x48 pixels and labels for different expressions
Summary
The Landmark-Driven Facial Expression Recognition system offers a novel approach to FER by integrating facial landmark detection algorithms. By avoiding the use of pretrained models and leveraging the accuracy enhancement provided by facial landmarks, this method demonstrates promising results on the FER2013 dataset. Researchers interested in exploring FER with a landmark-driven approach can utilize this system for experimental purposes.