Though this is the second semester we are working with Vyrill, our project with them involves a new problem statement and is not a continuation of last semester’s project. The project is centered around multimodal sentiment analysis. For a bit of background, sentiments are expressed linguistically, tonically and visually. This project involves the multimodal analysis of video images and audio for sentiment/emotion recognition. This involves performing audio and image analysis separately (using existing tools or by developing new ones) and then performing the joint multimodal analysis to detect sentiments and emotions. It also involves analyzing under what conditions which modality performs better and whether (or under what conditions) the multimodal analysis is more accurate.