Skip to content

Team · Team Project | Computer Vision and Recommendation Systems

Emotion Aware Music Recommendation

A multimodal affective computing experiment that captures facial emotion cues and maps user sentiment to personalized music recommendations — assistive personalisation, not clinical emotion diagnosis.

My contribution

I contributed to connecting emotion signals with recommendation flow and keeping the product narrative clear: human-centric interaction without overselling emotion AI.

Python · Computer vision · Facial emotion recognition · Machine learning · Recommendation systems · Affective computing

Problem

Traditional recommendation engines rely solely on historical listening logs and miss immediate shifts in a listener's real-time emotional state.

Solution & architecture

Combined a computer-vision facial expression recognition (FER) module with a mood-indexed recommendation engine, translating detected affect states into curated playlist selections.

Pipeline & UX

Face detection and alignment → emotion classification → sentiment-to-audio-feature mapping → track ranking. Inference intervals were designed to stay responsive without jarring playlist resets.

My contribution

Team contribution across CV–recommendation integration and product framing that stays honest about limits.

Impact

Demonstrates human-centric AI interaction by closing the loop between passive visual sentiment detection and dynamic content delivery.