Deep learning-based facial emotion recognition system

The system we created employs a convolutional neural network (CNN) to achieve facial emotion recognition. Our model received training from thousands of FER-2013 and CK+ images while we made it ready for deployment by converting images to grayscale and performing data augmentation through flipping and rotation. The model achieved successful emotion classification for six basic emotions. The system demonstrated high performance through its ability to handle challenging conditions such as low illumination and faint facial expressions while achieving 72.61% validation accuracy and 74.49% test accuracy. The research shows how this technology can be applied to real-time systems which include medical diagnosis and educational software that adapts to users.