Facial Emotion Recognition (FER) is a complex and crucial task in computer vision with widespread applications in fields such as human-computer interaction, psychology, healthcare, and marketing. Its goal is to accurately identify a person’s emotional state based on their facial expressions. However, the inherent variability of facial expressions across individuals, cultures, and contexts presents significant challenges. Recent advancements in deep learning, particularly in Convolutional Neural Networks (CNNs), have greatly enhanced the performance of FER systems, enabling near-human accuracy, especially in cases where emotions are clearly expressed. This thesis addresses FER in the context of multimodal datasets. The proposed method utilizes a CNN to extract salient features from facial images, coupled with a final transformer encoder and a classifier to predict the emotional state.