A consortium of European researchers coordinated by the Computer Vision Center (CVC) of the Autonomous University of Barcelona (UAB) has developed‘Hermes’, a system made up ofvideo cameras and softwarewho is able to recognize andanticipate human behavior, as well as describing it in natural language.
‘Hermes’ (acronym forHuman Expressive Representations of Motion and their Evalutation in Sequences) is based on the analysis of human behavior from video sequences captured with three levels of focus: that of the individual, as a relatively distant object; that of his body, with a medium proximity that allows to analyze his postures; and the face, which allows you to study your facial expressions in detail. The information obtained, processed by computer vision and artificial intelligence algorithms, allows the system tolearn and recognize movement patterns.
From the learned movements, ‘Hermes’ is able toanalyze and discover potential unusual behaviors and alert about them. For example, the system would send a warning to the control center of a subway station when capturing the image of an individual trying to get off the train track; or it would alert if an elderly person living alone suffers a fall. In addition to the many applications of the invention in the field ofsmart surveillanceand theaccident and crime prevention, the researchers consider that it also has great potential as a study tool in other fields, such as marketing or psychology.
The project, which has lasted three and a half years and has a budget of 2,100,000 euros, has had the participation of researchers from five of the most recognized research groups in Europe in this area and from the companyAnsware Technologies, specialized in information and communication technologies.
