Released face recognition software that analyzes facial expressions with an age error of only 5.27 years old

Released face recognition software that analyzes facial expressions with an age error of only 5.27 years old

Released face recognition software that analyzes facial expressions with an age error of only 5.27 years old

“Face recognition” is attracting attention as a marketing tool for companies. This is because the range of analysis will expand, such as what people are interested in and what kind of people are paying attention to them. Above all, it is becoming familiar software, so it seems that more companies will use it in the future.

Dynaflex Co., Ltd. announced on February 28 that it will start selling the face recognition software “DeepSight” from March 2020. This is software that can be used for marketing analysis such as digital signage.

No special camera required, real-time analysis possible

DeepSight achieves real-time facial analysis through computer vision and deep learning technology.

By analyzing image data, videos, and images taken by cameras, data such as gender, age, and facial expressions (six types such as happiness, sadness, and anger) can be obtained. Since data can be acquired in real time, it is possible to provide optimal digital content to the viewer. Of course, the collected data can be analyzed offline and used for planning marketing strategies.

The main things that DeepSight can analyze are:

  • age
  • It can be determined at +-5.27 years old relative to the actual age

  • sex
  • Can be determined in 93.0% of men and 91.8% of women

  • tracking people
  • Track passing customers and measure attention time.Even if you leave the field of view of the camera, those who return within a certain period of time will not be counted again.

  • head pose
  • Determine what your customers are paying attention to and what content is being displayed that attracts them and what they ignore

  • Counting the number of passers-by
  • You can count the number of people who pass by without facing the camera.

According to the press release, it is also possible to detect faces facing sideways, up and down, as well as faces with masks. Furthermore, it can be determined from 0 years old. Also, you don’t need a special camera or multiple cameras.

In addition, since the price is sold through an agency, it is necessary to inquire.

Face recognition technology used in vending machines

As mentioned at the beginning of this article, the use cases of face recognition technology are increasing. One of the things that has attracted a lot of attention recently is the vending machine.

JR East Water Business announced on February 13 that it will start demonstration tests of “multifunctional vending machines for inbound tourists” from March. Equipped with a camera equipped with a face recognition system, it is possible to identify attributes such as the user’s gender and age group. It will introduce recommended products according to the attributes identified, the outside temperature, and the time of day.

This vending machine was born from a survey of foreign visitors to Japan. It is said that Japanese vending machines give foreign visitors the impression that they are difficult to operate, that they do not know how to pay, and that they cannot distinguish between different types of beverages. Therefore, the vending machine used in the demonstration experiment is equipped with an AI that provides guidance and explanations in multiple languages.

In fact, I am personally interested in what kind of product is recommended, so I would like to use it when the demonstration experiment starts. Perhaps you can make a new discovery, such as being addicted to being recommended a drink that you don’t usually buy.

Related article: Are vending machines difficult for foreign visitors to Japan?