Samsung files a patent for Visual Recognition Technology

Samsung, a leading electronics manufacturer is working on an emotion recognition technology that is about to take the facial/visual recognition technology a step further, reveals a recent patent filing by the company on USPTO. Already, the facial recognition technology is well merged with most of the electronic gadgets from camera to mobile phones to identify the person in front of the imaging device. Those technology can identify faces, but not mature enough to identify the emotions on those faces. The below image represents how the AU's are recognized as emotions by their technology.

Table for emotion recognition from AU's

Samsung uses Facial Action Coding System (FACS), which are capable of describing facial expressions or emotions using Action Units (AU). Actually, AU's are nothing but the units representing the contraction of facial muscles. By certain, there are many facial muscle contraction need to take place for each emotion to be expressed by an individual.

The researchers Sudha Velusamy, Hariprasad Kannan, Balasubramanian Anand and Anshul Sharma who were working on Samsung's emotion recognition technology combined multiple AU's together that represent the emotion to form a single AU string. Then that AU string will be compared with a set of AU strings to have a best match for the emotion expressed by an individual.

The concept that Samsung puts forward is pretty impressive, but it needs an hardware apparatus to recognize the actual emotions expressed by an individual from AU's. Patent filing pretty much includes the hardware apparatus too and I believe this inclusion is to prevent the competitors from coming up with a similar technology anytime sooner.

The hardware for the emotion recognition apparatus includes input device, memory, processor, communication interface and at last an output device. The above components are pretty much available in a decent mobile phone itself, so anytime sooner we could expect a Samsung phone featured with emotion recognition technology.

Via: Electronista