Actlyzer

It defines and recognizes complex behaviors by combining about 100 pre-learned basic actions. "Using a smartphone while walking," for instance, can be defined and recognized as a combination of basic actions "walk," "face down," and "put one's arms forward." We are now developing a technology that can learn a complex series of actions from a single piece of training data.

We also develop technologies that utilize combinations of various characteristics of a person, such as how they are dressed and how they walk, to search for the person. It is also possible to estimate the height of a person, which is constant information, even from footage which is not very clear, using a technology to identify the height of a person from the video.

It has realized cost reduction for surveillance operations through optimizing the identification of suspicious persons and grasping their movement paths when an incident occurs by analyzing human behaviors and characteristics from camera footage on the premises.

It analyzes the facial expressions and the pulse of the person talking on the phone, using camera footage and information from the sensors. It then uses the analysis to estimate the psychological condition, such as lowered ability to judge things; this helps a person to avoid being taken in by phone scams, by visualizing risks. We are now conducting a field trial of phone scam avoidance in cooperation with the City of Amagasaki, Hyogo Prefecture.

Safe and secure store operation and higher-quality customer service has been realized by analyzing the behaviors of customers using in-store camera footage. This company has introduced Fujitsu's AI video analysis solution, including Actlyzer, aiming to convert its outlets into smart stores.