Several engineers are currently attempting to recognize the emotions of an individual based on facial expressions. Previously, the images used for human expression recognition had a high resolution and were of good quality. In practice, due to factors such as the camera sensor being sensitive to temperature and illumination and noise added to the transmission of images from a camera to a computer system, noise is added to pictures. It changes the pixel values. Past research has not studied the impact of these factors in great detail. We have proposed a Convolutional Neural Network (CNN) based human expression recognition system and evaluated its performance in the presence of different types of noise. We have also compared the CNN-based model with the Support Vector machine algorithm that has been used in the past to explain why CNN-based human expression recognition is more robust than other face recognition methods. Based on the findings from the experiments, we offer suggestions for developers using human expression recognition technologies.