A Haar-Cascading and Deep Learning Driven Enhanced Mechanism on Cloud Platform for Real-Time Facial Emotion Detection

Real-time emotion detection utilizes computer vision that aims to recognize and categorize emotions. However, it contains challenges, including the intricate nature of emotions and the variations in their expression across individuals. The accuracy of the system can also be affected by environmental factors such as background noise, lighting, and other people. Having access to a diverse and comprehensive dataset is essential for training algorithms that can accurately detect emotions in real time. In order to tackle these challenges, we utilized CNN and seven emotions to identify their corresponding facial expressions. By implementing appropriate activation functions like ELU, ReLU, and Leaky ReLU, and training with the dataset Facial Expression Recognition 2013 (FER-2013) that contains over 30,000+ images of facial expressions and emojis, as we increase the epoch the accuracy of this model increases, we achieved an impressive accuracy of 94.74% by using ReLU on 100 enochs.