Can Amazon really detect Fear by its Facial Recognition Technology?
Amazon’s Facial Recognition technology, “Rekognition” is a trending topic of 2019. On 13th August 2019, Amazon said that its facial recognition software can detect a person’s fear now. Rekognition is in talks these days because of the political tussle over the regulation of facial recognition. Though the United States regulatory agencies have deployed facial recognition; some municipalities (San Francisco, Somerville, Massachusetts) proactively opposed implementing this technology.
Amazon’s Rekognition has sparked interest among the audience because of its facial analysis and sentiment analysis for predicting emotions by identifying expressions from images of people’s faces. It has been accessing seven types of emotions: happy, sad, angry, disgusted, surprised, calm, and confused. And now, “fear” is the eighth.
Know that Amazon isn’t the first company to offer emotion detection algorithms. Back in 2015, Microsoft introduced similar services that look out for emotions, however, they added “contempt” instead of “confusion.” Then, the very next year Google also offered its services that claimed to detect emotions. However, it remains an open question that can Amazon’s Rekognition be the first one to successfully incorporate the concept of detecting fear through a person’s face image or video?
Though Amazon hasn’t been expressive about the usage of Rekognition, however, they are clear regarding the provision of services to customers working in the police department. Also, on the Rekognitikon website, Amazon suggested that e-commerce stores can use this technology to analyze and track the demographical and emotional trends of customers at different store locations.
However, psychologists are skeptical about reading emotions solely from facial expressions.
According to a study published in 2018 by the University of California, Berkley, to accurately read the emotions of a person from video, it is essential to pay attention to body language as well.
It undermines the core assumptions of emotion-detection software; i.e. detecting emotions from facial expressions only.
The online documentation released by Amazon regarding Rekognition clearly says that the API only analyzes the physical appearance of a person’s face to detect the emotion. It doesn’t determine the internal state of a person at the time of detecting emotion. What if they are faking expressions to deceive the technology?
Considering different psychological facts and researches, the current version of Rekognition seems vague regarding the accuracy of emotion detection. Not just the fear, but through emotion-detection algorithms, it can be challenging for Amazon to decipher the emotions behind facial movements. Similar facial movements can indicate more than one category of emotions. Also, the use of facial recognition has raised concerns regarding privacy.
