Facial emotion detection, a pivotal component of Artificial Intelligence (AI) and Computer Vision (CV), aims to recognize and identify human emotions from facial expressions. This paper presents an approach leveraging the Deep Learning (DL) models that includes Convolution Neural Network (CNN), Dual-Temporal Scale Convolutional Neural Networks (DTSCNN), Recurrent Neural Networks (RNN), and Residual Networks (ResNet-50) to achieve real-time and accurate emotion recognition. The primary objectives encompass real-time emotion recognition, high accuracy, low latency, and robustness to varied conditions. This paper performs experiments on benchmark datasets for evaluating each model considering accuracy, processing speed and facial orientations. This paper highlights the study of comparison between these models. The outcomes indicate that the CNN approach outperforms other methods, yielding superior accuracy and robustness. This research contributes in the advancement of facial emotion detection, with implications for applications in human-computer interaction, psychology, marketing, and healthcare. The CNN ensemble model represents a significant advancement in facial emotion detection, offering a comprehensive solution with broad implications across diverse domains. Its effectiveness highlights the significance of continuous exploration and refinement of deep-learning techniques to address complex tasks in CV and AI effectively.

Deep Learning Emotion Detection CNN RNN DTSCNN ResNet-50