Tensor Facial Recognition Technology Partners Herta Security have launched a new emotion detection solution that can be used to analyse and categorize people’s facial expressions in video footage. Dubbed BioObserver, the solution was built with input from enforcement professionals, who will also serve as some of Herta’s primary customers.
According to Herta, BioObserver is a facial analysis solution for the automatic detection and annotation of a person’s emotional states and micro-expressions.
It is a non-invasive software for the individual, since it is based on image processing techniques. It is able to detect basic facial emotions such as “joy”, “sadness” or “anger”, and also more subtle micro-expressions of the face such as “frown”, “blink” or “eyebrows raise”.
BioObserver also allows to extract the direction of the gaze and the orientation of the head, to monitor behavioural metrics such as the degree of attention of the individual. In addition to automatically tagging the extracted facial information, the platform allows you to configure additional annotations of events and frames that are considered of interest.
In the context of an interview or interrogation, the analysis of facial expressions can provide invaluable support to the observer. The spotter can assess, for example, in which specific moments they occur in relation to the question presented: when listening to it, while processing that information; when answering, after having given the answer.
It is also interesting for the detection of emotional incongruities, that is, situations in which the subject verbally expresses an emotion while showing a very different one on the face. Likewise, the direction of the gaze and the orientation of the head over time express the degree of attention of the interviewee, giving clues about their interest, abilities and certain personality traits.
BioObserver analyzes the face frame by frame, either from a pre-recorded video or from a camera capture in real time. It begins by detecting the presence and location of the face within the frame. Next, it extracts a series of characteristic points within the face (for example, around the eyes, eyebrows, nose and mouth), with various purposes.
First, to determine the orientation of the gaze and the head. Second, to trim and align the facial region. It is this cropped and aligned image of the face that is finally used for classification in terms of basic emotions, micro-expressions and behavioral metrics.
BioObserver’s classification algorithms are based on Deep Learning, an advanced Artificial Intelligence technique that uses deep neural networks. These algorithms are capable of automatically extracting the most relevant information from the face, such as patterns and textures (for example, presence of wrinkles around the eyes, shape of the mouth, etc). They have been trained with an extensive database of millions of images of subjects of different ages, genders and ethnicities. This allows BioObserver to maintain a robust and universal behavior, with very high hit rates.
Tensor – experts in deploying high-end facial recognition systems
Tensor’s facial recognition technology is simply – the best there is. It’s designed specifically for fast, discrete identification/registration of known/unknown individuals, whether in a large crowd at a stadium or festival or in public places such as airports, train stations, shopping centres or urban areas.
The system dynamically compares images of individuals from incoming video streams against those stored in a predefined list and immediately alerts the appropriate people when a positive match occurs. The technology yields a very good level of performance even when there are partial occlusions of the face; the use of glasses, scarves or caps, changes of facial expression and moderate rotations of the face. Moreover, it does not allow users to be impersonated using photographs.
Our Facial Recognition CCTV system is easily integrated with existing ID management products and enables administrators to create Whitelists/ Blacklists for specific areas. People are easily and quickly enrolled from one or more photographs and/or video clips, negating the need for a subject to be physically enrolled.
Optimized for GPU computing architectures, the Tensor Facial Recognition system is 40 times faster than traditional systems (up to 150 fps), and works with very high-resolution video streams – up to 150 fps with high resolution IP cameras. It analyses multiple cameras simultaneously and fully supports runtime alarm management, is highly tunable (control based on time-frames, sequentially, etc.) and even allows alarms to be exported to common formats (PDF, Excel) and remote devices (mobile, PDA, tablet, security control centre).