A Web Application to Recommend Songs Based on Human Facial Expressions and Emotions

Facial expressions are a common non-verbal way of how humans show and express their emotions to others. Emotions can be categorized as positive and negative emotions, derived from facial expressions, in which negative emotions can affect a person�s behavior and thinking. Music is a common remedy for people to cope with both positive and negative emotions. The use of deep learning to identify emotions based on human expression can be an effective and efficient way to provide solutions for humans because it can mimic the way humans think while requiring less time and effort. By creating a solution that encompasses recommending songs from emotion detected using deep learning, it can benefit society health and entertainment-wise. This paper presents a project that focuses on developing such a solution and testing its performance and effectiveness to users, in improving their emotions via songs. The method used for this web application is OpenCV and DeepFace as face detector and emotion recognition system, respectively; while the song recommendations are pulled via Spotify API, where all these elements are deployed in a web application using Streamlit. DeepFace has been stated to have an accuracy of around 97 for its facial recognition functionality, along with their facial attribute analysis, which can be considered reliable enough to recognize emotions. For future work, other factors that can help to identify emotions are to be put more focus on, as it is envisaged to improve the emotion recognition system in this web application. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.