Herta Security has launched a new video analysis solution, BioObserver, to provide a platform for automatic micro-expression detection based on face biometrics. The biometric solution is the result of a collaboration between Herta and key end-customers in the law enforcement sector, according to a company announcement.
AI-based BioObserver, aimed at interviewers, can analyse 7 basic facial emotions, as well as 18 micro-expressions, such as ‘frowning,’ ‘blinking’ or ‘raising eyebrows,’ according to Herta’s website. The combination of various expressions can be collated into a higher-level prototypic facial emotion, and the platform is able to intuitively detect the direction of the gaze and head orientation. This helps to create an efficient and easy-use tool to monitor behavioral metrics in an individual, the announcement states.
However, previous research has found that a common training data issue in facial recognition algorithms have also been found in expression recognition technologies. Scientists found expression recognition algorithms cannot be made to perform evenly between demographics due to the lack of diverse data in the biometric databases they are trained on.
Herta’s platform allows users to configure automatic rules and annotations of events to extract keyframes of the video that are considered of interest. BioObserver is being offered via cloud-based software, enabling customers to rapidly process videos in a short amount of time. In 2020, the company was awarded a COVID-19 Response Seal of Excellence by the European Commission for a biometric crowd-monitoring system, which answered a call for innovative technologies at the beginning of the pandemic.