Emotion Detection and Sentiment Analysis of Static Images

Abstract

The usage of social media platform such as Facebook, Instagram, Flicker, etc. is rising day by day wherein images play a major role. It is said “An image is worth a thousand words”, people these days upload certain images on these sites to display their sentiments and emotions in the form of picture on almost every occasion. Images play the most important role in today's generation where it has become a major part of everyone's lives. Most of the prevailing research have focused on sentiment analyses of textual data, but only limited researches have focused on analyzing sentiment of visual data. In this project, we have explored the possibilities of Convolutional Neural Networks (CNN) to predict the various emotions (happiness, surprise, sadness, fear, anger and neutral) depicted by an image. These sort of predictions can be useful in applications for automatic tag predictions of the visual data available on social media platforms and understanding sentiments of the people and their emotions.