Fujitsu: AI-system for measuring attention concentration

2021: Announcement of AI-system of concentration measurement for employee control

In mid-March 2021, Fujitsu introduced an AI system that can quantify the concentration of people in performing tasks. According to Fujitsu developers, the new AI system estimates a person's concentration by highlighting the general characteristics of the face and analyzing each muscle group separately. This avoids errors related to the cultural characteristics of the subjects and the characteristics of the tasks.

The technology captures brief changes in facial expressions lasting only a few seconds, for example, pressed lips, as well as long-term changes, such as a close look, and evaluates them relative to optimized time frames, the developers explained.

Results of evaluation of human concentration by AI solution from Fujitsu

To create an AI algorithm, Fujitsu collected data from 650 people from China, Japan and the United States as part of the collection of data for machine learning. The database included people whose work is associated with increased concentration of attention, such as memorizing and searching for information. According to the developers, the new system allows you to estimate the degree of concentration with an accuracy of more than 85%. Fujitsu also notes that the system is able to detect a decrease in concentration due to drowsiness.

Fujitsu hopes that the new system will find its application in various services providing training and meetings on the network. Meanwhile, in early 2021, the Council of Europe published new recommendations criticizing the use of facial recognition technologies in the workplace and in the educational environment. The Council of Europe intends to prohibit the use of digital tools that can recognize emotions, identify personality traits or mental states, since they can be used in recruitment, insurance and education, leading to discrimination of people.^[1]

Notes