CNN Based Unimodal for Facial Emotion Recognition
Abstract
ER plays an important role in improving HCI systems by enabling computer systems to detect and understand human emotions. In this study, a deep learning-based FER system is proposed which uses a pre-trained ResNet-50 Convolutional Neural Network model. The model is fine-tuned on CK+48 facial expression dataset to identify the seven basic emotions. While many models achieve high offline accuracy, they often fail in real-time HCI applications due to high inter-frame flickering and processing time. Instead of a completely isolated architectural novelty, this study develops an end-to-end lightweight pipeline designed to meet the demanding response time needs of real-time Human-Computer Interaction (HCI). Our model successfully bridges the gap between the high offline feature extraction and flicker-free, low latency visual feedback. The system uses a Haar Cascade frontend and a fine-tuned ResNet-50 backend and consists of a deterministic temporal stabilization buffer.