Comparative Study on Sentiment Analysis in Image-Based Memes

In this electronic age of computers and mobile devices, users frequently share and express their opinions through videos and images. Memes are shared as a means of expressing the user’s thoughts, feelings, experiences, and sentiments. Memes are being used extensively, which has both pros and cons. Alarmingly more offensive and sensitive memes are being shared, which may be upsetting to other people’s social and cultural values. In order to classify these memes as positive or negative, this study will investigate their meaning. Thus, it will allow us to comprehend the user’s opinions on many topics and concepts. We have deployed different multimodal and unimodal on Facebook’s dataset of hateful memes and categorized the memes into positive and negative types. The best modal found was $BERT$, which outperformed all other algorithms with an accuracy of 93%.