Facial emotion recognition (FER) is a method for detecting emotions via facial expressions. By means of this technique, computers or robots can understand human feelings. FER is a modern invention which may aid in identifying the human emotions from facial expressions. Before the extensive use of CNN, FER trusted many methods, like (SVM) support vector machines. FER has an inspiring task to compute human interactions. FER got significant attention in past few years. Facial emotions are crucial aspects of human connections that help us acknowledge the purpose of others. Analysts in this area are focused on generating different approaches to explain facial expressions and bring out certain features to have a better projection by computer. Because of the exceptional success of deep learning, various types of architectures are utilized to attain good performance. In this paper, we develop a CNN model for FER via deep learning. Here, a sequential model is used for this work, "reLU" is used as an activation function, and "adam" as an optimizer. This model is trained using the FER-2013 dataset and achieved an accuracy of 84.21% with just twenty epochs.