Startup Sensing Feeling offers video emotion analysis without facial recognition
A UK startup is using computer vision with facial and body language analytics to assess the emotions of people and crowds in real time for risk-assessment applications, the Evening Standard reports.
Sensing Feeling can interpret anger, contempt, fear, disgust, happiness, sadness, and surprise, and could have use in industrial, retail, educational, arts and office settings, according to the report. Staff could be alerted to tired workers, upset customers, or frustrated crowds with the technology, which is already used to provide student satisfaction feedback at a university outside of the UK.
“Our technology can sense how people are feeling, just by observing their behaviour,” Sensing Feeling founder Jag Minhas told the Standard. “It can assist in the more effective management of that by highlighting risk associated with crowding in public spaces, or behaviour that might cause accidents and near misses. A pattern of behaviour that might be associated with the risk of self-harm could also be something that this type of solution can detect.”
Minhas also told the publication that his company recognizes privacy concerns with computer vision technology, and has safeguards in place to ensure data regulation compliance. The Standard reports the company’s patented technology maps facial characteristics, hand gestures, upper and lower body pose, and interactions with other people to recognize behavioral patterns, and return an “emotional index” score.
Sensing Feeling’s deep learning algorithms deliver behavior detection with 95.76 percent accuracy against reference datasets, 74 times faster than state of the art models, and requires low levels of computing power appropriate to low-cost edge deployments, according to the company’s website. The technology is also advertised as GDPR-compliant out of the box. It does not record images, use facial recognition, or tracking.
The company was started with a £72 thousand (US$94 thousand) grant from Innovate UK and has raised £436,500 (roughly $571,000), according to CrunchBase, and is supported by tech accelerator Wayra UK.
Affect recognition is a booming market despite indications that the methodology behind it may be unreliable.