This Master’s Thesis aims to design and develop a functional smart mirror prototype capable of detecting and responding to the user’s emotional state in real time using artificial intelligence techniques. The project builds upon previous solutions by replacing a Streamlit-based interface with a modern QML-based graphical interface, better suited for embedded environments and continuous interactive systems. The proposed system integrates automatic facial emotion recognition through computer vision and deep learning techniques, relying on libraries such as OpenCV and DeepFace for emotion detection and classification. In addition, the system architecture has been designed to support multimodal reasoning, allowing the combination of visual and contextual information to improve recognition robustness. Detected emotional states are presented through an adaptive graphical interface that dynamically adjusts its visual behavior according to the user’s emotions. Within this framework, the feasibility of real-time video processing on resource-constrained hardware is analyzed, evaluating model performance, system latency, and user interface responsiveness. Optimization strategies are applied to ensure smooth and continuous interaction. The system achieves over 85% accuracy in the classification of basic emotions under controlled conditions, demonstrating the technical viability of the proposed approach. Finally, this thesis discusses the limitations of embedded systems for deploying emotionaware applications, as well as the social and ethical implications associated with emotion recognition technologies. The work contributes to the development of more efficient, adaptive, and user-centered interfaces for intelligent devices and IoT applications, particularly in contexts related to emotional well-being and human–computer interaction.