Enhanced Facial Expression Recognition: A Convolutional Neural Network Approach

Sumit Sanjay Maske Independent Researcher, Bloomington, Indiana, United States.

Avalokiteshvara Journal of Artificial Intelligence (AJAI)
Volume 1, Issue 1, March 2025, pp. 19-25
Research Article

---

Abstract: Within non-verbal communication, the identification of facial expressions stands as a significant and formidable task. The objective of a facial expression recognition system is to categorize real-time facial images into distinct emotional classes, encompassing Anger, Disgust, Fear, Happy, Sad, Surprise, and Neutral. Drawing inspiration from the advancements achieved in image recognition and classification through Convolutional Neural Networks (CNNs), this paper advocates for a CNN-based methodology to tackle the complexities of facial expression recognition. The model presented in this study harnesses various libraries, including OpenCV, Keras, and TensorFlow. Through the utilization of grayscale images from the Face Expression Recognition dataset on Kaggle, the model undergoes training with CNN architectures of varying depths. Notably, the proposed model attains an accuracy of 72.34% for familiar data and 60.54% for previously unseen data.

Keywords: Facial Expression Recognition System, Deep Learning, Convolutional Neural Network, OpenCV

📄 Download Full PDF