By Thale Kirkhorn, Svenja Lys Forstner and Kaja Hustad Bendiksvoll
In the project, we focus on a support solution for a condition called “social-emotional agnosia”, also known as “social blindness” (mdmedicine, 2011). It describes the inability to interpret facial expressions, body language, and tone of voice. Individuals with this condition experience difficulties with accurately understanding another person’s emotions in social situations. Picking up non-verbal communication during conversations is especially challenging. A device such as WIME could assist in social situations and improve the flow of the conversation.
The emotion recognition system uses a pre-trained machine learning model to recognize facial expressions from a real-time video stream. The software detects the emotion of one person at a time. At the moment, six different emotions are recognized: happy, angry, sad, surprise, fear, and neutral.
When an emotion is detected a file is created with the data of the current emotion. The file updates each time a new emotion is recognized. The file is hosted on a local webserver, where the smartwatch is connected via bluetooth to retrieve the emotion file.
For the visual feedback, we use the programmable smartwatch “Bangle.js” by Espruino. For each of the emotions, we created a different screen output. Six different emojis represent one emotion each. A text is also
displayed underneath, to further clarify the emotion detected.
See the whole presentation > here <
