Emotion Detection
A single label cannot capture non-basic complex expressions with variations in emotions or intensity.
The model holds that all emotions can be described as a linear combination of two underlying neurophysiological systems (Arousal and Valence), with all emotions shading imperceptibly from one into another along the contour of the two-dimensional circumplex.
The technology addresses major issues in the current emotion recognition. We do not use prototypical models, but rather, a psychologically-plausible dimensional analysis across the Arousal and Valence attributes:
Arousal is an attribute that describes how energetic or not an emotion/expression is
Valence on the other hand describes the how positive or negative an emotion/expression is
A wide range of emotions can be represented with different Arousal/Valence values
It is far superior than 7 basic labellings (surprised, sad, afraid, neutral, disgusted, angry, happy) which lack contextualisation as it is a biased selection of attributes with stereotypes and prejudices ingrained using pictures or videos for training.
Data pertaining to trends of emotional attributes over time and aggregated statistics is displayed in real-time.
For groups, we provide comparative analytics to identify how different people react to different situations.
For larger crowds, we measure and extract aggregated statistics from anyone involved.
