SyncPixel: A Computer Vision Framework for Automated Emotion-Based Music Suggestions | IJCT Volume 12 – Issue 6 | IJCT-V12I6P19

This study presents an innovative deep learning–based framework that bridges the emotional gap between visual perception and auditory experience through an intelligent, image-driven music recommendation system. By integrating advanced computer vision and affective computing techniques, the proposed model successfully interprets the emotional context embedded in visual content, translating it into meaningful musical associations. Through the combined use of DeepFace for facial emotion recognition and BLIP for image captioning and sentiment extraction, the system captures both explicit and implicit emotional cues, including facial expressions, environmental context, lighting, and color tone. These multimodal features form the foundation for accurate emotion detection, which is then mapped to appropriate musical parameters such as energy, valence, and tempo via the Spotify API to generate personalized, emotion-aligned playlists. Experimental evaluation demonstrates that the system is capable of effectively correlating visual emotions with music genres and emotions, offering users a more intuitive, emotionally resonant, and context-aware listening experience. The approach reduces decision fatigue in music selection while fostering emotional engagement and digital self-expression. Furthermore, the use of a Streamlit-based interface ensures seamless interaction and accessibility, allowing real-time processing and visualization of detected emotions alongside the curated playlist. From a research perspective, this work contributes to the growing domain of affective computing, human–computer interaction, and multimodal recommendation systems by showcasing how emotion understanding from non-verbal cues can enhance content personalization. The model demonstrates scalability and adaptability, making it suitable for integration into social media platforms, content creation tools, and emotion-driven entertainment systems. Future research directions include enhancing the emotion detection pipeline using multimodal fusion with audio and textual data, implementing real-time feedback mechanisms to refine the emotion–genre mapping dynamically, and exploring transformer-based architectures for deeper contextual emotion understanding. Expanding the emotional taxonomy beyond basic categories and incorporating cross-cultural emotion modeling could further improve global applicability. In conclusion, the proposed framework lays the foundation for next-generation emotion-aware music recommendation systems that not only respond to user preferences but also resonate with their psychological and emotional states. By merging artificial intelligence, human emotion, and artistic expression, this research represents a meaningful step toward more empathetic, adaptive, and human-centric digital ecosystems.