Facial emotion recognition (FER) is very important in determining emotions and subsequently predicting the behavioral patterns of humans, with respect to social interactions. Due to the capability of FER systems in recognizing the true emotional states on human faces, the system has wide application areas, including, psychology, HCI, and healthcare. Currently, the FER systems have experienced a boost in their level of accuracy and automation of the process with the incorporation of machine learning models, making it possible to provide the FER reports on the real-time basis almost entirely. In this paper, there presented is a novel approach on FER using Machine learning that concentrates on Convolution neural networks for image processing and emotion categorization. The proposed method is more accurate and fast compared to conventional approaches to emotion recognition and thus can be of high practical value in various applications including emotion-sensitive technology in social robotic design, health management, and individual learning. The results established that the proposed system given their performance, has a very good efficiency in recognizing emotions, even in low light and with faces occlusions. This paper discusses the research opportunities provided by FER systems to enhance the knowledge of the social processes and social relations between people and technology.