Our Emotional Reaction model uses AI to detect the emotion of your customers by analysing their facial expressions while they interact with your product.
With Emotional Reactions you are empowered with deeper insights about your customers and how they really feel about your product.
How the model works
Based on a calculated baseline of a range of different emotions, the model can detect deviations in emotion and correlate peaks of emotions to the stimulus that your customer is looking at. If an emotion is detected by the model, a representative emoji is attached to an observation nugget, which is then used for further analysis. As a user, you are also able to delete an emotion from a an observation or even change it to another emotion that you think is more fitting. Our AI models are continuously getting smarter based on user input, and by changing the emotion, you also help the model to learn from its mistakes and improve its accuracy in the future.
The Emotional Reaction model was built and trained on 30.000 image-examples and achieves 95% accuracy when tested on CK+, a widely used benchmark database for facial expressions.
The model sums all deviations from the baseline for each participant and divides them by the number of observational nuggets, which gives an average deviation per emotion.
Emotion over time
The model aggregates the predicted emotions per second of all observational nuggets in the insight and calculates the amount of time each predicted emotion was peaking.