Facial expressions serve as crucial social communication tools, offering significant insights into people’s mental states. To accurately extract this information, extensive research is being conducted, especially in the area of human–machine interaction. The importance of research in the area is growing. Numerous methods have been developed to recognize distinct facial expressions—neutral, happy, surprise, sad, anger, disgust, fear, and neutral with the use of intelligent systems with high accuracy. However, emotion recognition remains a challenging task because of some factors like insufficient light, age, race, gender, and facial position. The presented paper considers two-stage model for detecting emotions from images including face. In the initial stage, a deep learning-based CNN model is utilized for feature extraction from the FER-2013 dataset. Following feature extraction, classifiers such as Logistic Regression (LR) and Random Forest (RF) are employed for classification. The same dataset is used to test the proposed model, yielding a classification accuracy of 48% with RF and 58% with LR. Additionally, the paper describes a CNN model that handles both feature extraction and classification, achieving 70% accuracy on the challenging FER-2013 dataset.