Facial Emotion Recognition (FER) enables automatic classification detection of human emotions from facial expressions using deep learning (DL) and computer vision techniques. In this paper, a hybrid real-time emotion recognition system using Convolutional Neural Networks (CNNs), OpenCV, and DeepFace is proposed to achieve accurate and dynamic emotion analysis. The system aims to identify emotions such as happiness, sadness, anger, surprise, and fear with a high level of accuracy to enable enhanced Human-Computer Interaction (HCI), mental health monitoring, and smart safety systems. The method used includes offline training of the model and real-time emotion recognition using video. The technique employs continuous learning and optimization strategies to maximize recognition rates and resilience in practical environments.