Analyzing Expressions to Curate Feeds: A Real-Time Mood-Based Reels Algorithm

As technology evolves, the demand for customised and personalised experiences is increasing in daily life. This tendency is primarily noticeable in the field of social media platforms, where user satisfaction is pivoted on the content that relates well with the individual’s preferences and interests. Meeting these demands requires emphasis on understanding and addressing user-specific needs and requirements. Vertical videos and reels are rising in popularity and growing rapidly nowadays at a faster rate due to their new style of representation. A huge random cluster of reels makes it challenging for users to find interesting and relevant reels that align with their current mood hence to keep them engaged on the platform. To overcome this lack, this research presents a solution, namely Reel Mood, an innovative algorithm that employs the latest facial detection technology with Flutter. It has been designed as an initiative that will reshape recommendation algorithms. The research detects the user’s emotion through facial expressions using the device’s camera and captures images throughout. From those images, their mood is analysed, and based on their current emotion, our Reel Mood application recommends reels tailored as per their mood. The facial expression machine learning model integrates smoothly into the Flutter application, featuring an engaging user interface: recognising facial expressions and recommending reels based on their current mood.