Facial Expression Recognition with CNN-SVM for Emotional State Classification

Facial expression recognition is useful for many things, including human-computer interaction and emotion analysis. For the purpose of recognizing different facial expressions, we present a Convolutional Neural Network (CNN) and a Support Vector Machine (SVM) classifier in this research. Using a large-scale dataset, experiments confirm that the proposed system is effective at using the CNN-SVM model for recognizing seven distinct facial emotions (neutral, angry, surprise, fear, sorrow, disgust, and happy). The experimental findings were very reliable, with high F1 scores and excellent precision across a variety of memory and precision metrics. All classes of facial expressions are accurately interpreted by the algorithm, although the ‘Angry,’ ‘Fear,’ and ‘Happy’ classes do very well. The CNN-SVM model demonstrated its viability in real-world settings by achieving an overall accuracy of 98.06%. With an F1 score of over 97.80 percent, it's clear that the system can recognize a broad variety of emotions. Emotion identification in HCI, sentiment analysis, and content recommendation systems are just few of the areas where these findings highlight the value of the CNN-SVM paradigm.