How do people really react to YouTube videos? Which parts of videos are most interesting?
Tools: node.js, Python (Django), mongodb, EMOTIENT emotion recognition
This was an exploration of Emotient facial emotion recognition technology (later acquired by Apple).
- How well does the technology work?
- How comfortable are people with the emotion tracking?
- What types of activities are the most emotionally charging and how well can they be captured by this technology?
- What could this information be used for?
So I came up with the idea of Emotional Web that will measure people’s reactions to YouTube videos. As people watched various videos on YouTube, their facial reactions were analyzed and could later be used to aggregate the most significant moments from various videos, or even compare them according to certain emotions. One could use this to create better recommendation systems, review your emotions over time, perhaps correlate them with other activities as a part of quantified-self movement etc. I built this to be deployed internally within our group at Intel to do a longitudinal study of these questions. Funniest cat videos, anyone?