Tune Mood Emotional Recognition for Personalized Music

Emotion recognition systems have gained significant attention due to their ability to understand human feelings through visual cues and provide personalized responses. This study presents “Tune Mood: Emotional Recognition for Personalized Music,” a web-based application that detects human emotions from uploaded images and recommends music according to the detected mood. The system integrates deep learning techniques to perform accurate facial emotion analysis. For image-based emotion prediction, the MobileNet model is utilized due to its lightweight architecture and strong performance in facial expression classification tasks. To enhance the reliability of emotion detection, the system also incorporates a deepfake verification step that analyzes whether the uploaded image might be fake or edited before performing emotion recognition. Facial features are analyzed using the DeepFace library to extract emotional attributes from detected faces. The application is implemented using Flask, OpenCV, and MySQL, enabling user authentication, image uploading, and emotion analysis through a web interface. The system identifies multiple emotional states such as happiness, sadness, anger, surprise, fear, disgust, and neutral expressions. Experimental results demonstrate that the MobileNet model achieves high accuracy for image-based emotion recognition while the deepfake verification step helps ensure that the analyzed images are authentic. By combining emotion recognition with music recommendation, the proposed system enhances user experience and provides mood-based personalized entertainment.

Keywords

Emotion Recognition, MobileNet, DeepFace, Deepfake Detection, Facial Expression Analysis, Personalized Music Recommendation, Deep Learning, Computer Vision.