Views: 245 Author: Site Editor Publish Time: 2020-12-05 Origin: Site
The impact of the epidemic in 2020 on the society is earth shaking. One of the most obvious changes in life is that people's faces have new “decorations”- face masks. People wear various types of masks made of different mask machines according to their own needs, N95 masks with strong defense, and personalized masks that pursue fashion and medical functions. Various masks not only extend the distance between people, but also make today's intelligent life degenerate. The widely used face recognition technology needs further iteration due to the appearance of mask.
In the "face brushing era", face recognition with masks has become a big problem.
The traditional face recognition algorithm has been unable to adapt to the situation that the face is covered by a large area. This is mainly due to the following conditions.
1. A mask covers half of the face and directly loses a lot of facial features.
2. A large number of mask wearing face images can not be collected in a short time, so the algorithm training is more difficult
3. Face recognition system includes detection, tracking, recognition and other modules, which will affect them.
Facing the rapid spread of coronavirus disease in the world, even breaking out in the communities and big cities. To confront with the severe epidemic situation, many companies have carried out the research and application of fast analytical technology. For example, Baidu in China has released a new model with multi-functional detection which is able to recognize human faces.
In face recognition, the extraction of facial features is the most critical step, but now people wear masks made of multi-layer non-woven fabrics or self-made cloth masks. How to recognize faces through the surrounding masks or other methods is a big problem.
Eye feature is the key, through the spatial location-based attention mechanism feature learning, recognition algorithm can obtain more features about people's eyes, in this way, the loss of information can be minimized.
At the same time, combined with temperature sensing technology, face recognition temperature technology came into being.
Face recognition thermometers can be widely used, such as residential areas, office buildings, industrial areas, subways, bus stations and other occasions that require personnel access management and temperature measurement. The installation methods include wall-mounted face recognition machines or channel gates to achieve personnel Identification of entry and exit, statistical analysis of personnel data, management of temperature detection records, etc., can be set according to the number of users to enter the device, channel management, effectively reduce the detection workload, while reducing the risk of cross infection.
However, the premise of the application of face temperature recognition technology is the self-improvement of people's awareness of wearing masks. As a mask supplier, not only need to provide high-quality masks, but also should learn how to teach people to wear masks.