Face detection has increasingly become a cornerstone of advanced video analytics. Face detection is very different from, and should not be confused with, facial recognition. Face detection algorithms look for the typical features and characteristics of a human face, allowing a high degree of certainty that the video includes an actual person rather than another object which may be similar in size, shape or behaviour to a person.
In contrast, facial recognition uses face detection to identify the presence of a person, but then compares the facial information to a database of known subjects to find a match. Real-world trials have shown some difficulties in achieving a high level of accuracy with facial recognition, to say nothing of the privacy issues. Face detection doesn’t have these problems, and is predominantly used to filter events or trigger secondary analytics.
Face detection was once reliant on mapping a number of characteristics of the human face, but with the advanced use of artificial intelligence and machine learning, the accuracy has significantly increased, even where parts of the face may be obscured. This has ensured that face detection remains credible and effective during the pandemic, as well implemented algorithms can accurately detect faces, even in the person is wearing a face mask.
Indeed, this capability has proven invaluable for many sites as they return to working and need to meet regulations associated with the ‘new normal’. A number of sites including commercial offices, retail stores, hospitality sites, leisure facilities, transportation and public amenities are reliant upon the implementation of face mask policies in order to operate safely. Because high-level smart face detection can detect people even if they are wearing a mask, this allows the algorithm to quickly and accurately detect people and create an alert based upon whether or not they meet requirements to wear a face mask in specified areas.
A.I. Tech has developed AI-FACEDETECT-DEEP, a video analysis solution designed to assist in the implementation of regulations to tackle the coronavirus pandemic. In helping businesses and organisations to prepare for the new normal, it not only ensures the safety of customers, staff and contractors, but also delivers more efficient business practices.
AI-FACEDETECT-DEEP is a new video analytics application which can detect faces within a defined area, generating a notification if people are (or are not) wearing a face mask. Installation and set-up is simple, as the algorithm only requires the camera to be positioned with a frontal view at the height of an average person. It can deliver consistent and accurate results whether deployed in indoor or outdoor environments.
The application makes use of advanced traditional machine learning and deep learning algorithms, performing both face detection and face analysis to verify circumstances, based upon user requirements.
AI-FACEDETECT-DEEP is a perfect video analysis solution for locations where it is essential to verify those who are accessing the site are wearing a face mask, such as shops, banks, gyms and leisure facilities, offices, universities and schools, stations and airports.
AI-FACEDETECT-DEEP is a multiplatform solution, as is the case with many other A.I. Tech applications. It is supported by more than 20 different hardware platforms. It can be loaded as an edge-based application onto cameras and encoders from a number of manufacturers including Axis, Dahua, Hanwha Techwin, Hikvision, Mobotix, Panasonic and any cameras which support the Security & Safety Things OS, or deployed as a server-based solution. It is also available as an embedded application on the AI-Appliance-Deep (A.I. Tech’s own embedded server solution) which delivers an integrated hardware and software option powered with GPUs from NVIDIA.
