The reason why F-22 Raptor is called the “stealth fighter” is not because it can disappear on the radar screen (and it can’t) but because it can greatly reduce the recognition rate of mainstream radars. Therefore, it can come and go without leaving a trace. Our anti-facial recognition privacy glasses are the same. They reduce the facial capture success rate of most of the current monitoring cameras and disturb the algorithm to reduce the recognition rate. In this way, the infrared monitoring camera fails to capture the facial information and cannot carry out the correct matching with the data in the background database.
For preventing being recognized by facial recognition machines in a certain place or period , you may use a scarf or a mask and hide your face in a “safety box”. In fact, most people live in the cities with 630,000 cameras like London, supermarket and urban street. Even in the open and deserted field, you may under the surveillance of UAVs hovering in the sky. You are exposed to monitoring cameras all the time. Therefore, you can not guarantee to put on the “anti-facial recognition safety box” before entering the place with camera monitoring or carrying a stupid anti-facial recognition toy all day long when you dressed up in exquisite clothes of high-end brand. You are invisible in the AI world while becoming the eye-catching monster or clown in the reality. Therefore, the best choice should be protecting our facial information privacy without being noticed by others, while integrating the protecting equipment into our daily live and can match the personal image anytime anywhere simultaneously.
For 2D facial recognition systems in the early days, if the face of the subject is intentionally covered, for example, by fake beard, heavy makeup and sunglasses, it is very difficult to recognize the subject. The facial recognition is evolving. At present, 3D facial recognition system can capture the facial information through infrared ray or infrared ray laser penetrating the sunglasses for the face with sunglasses and mask. In this way, the facial information behind the sunglasses shows the real feature in the infrared ray camera.
Our anti-facial recognition privacy glasses adopt the unique optical means and successfully intercept the transmission of some visible light, infrared light and infrared ray laser, and disturb the camera’s collection and detection of eyes and surrounding key facial information. Due to the lack of key information, the recognition rate of facial detection system declines or even fails. Besides, the database’s correct record of new facial information can be disturbed. The research and development are conducted based on the facial recognition technology principle of mainstream monitoring camera.
Japan has developed an anti-facial recognition glasses with near infrared LED lamp. The light is invisible for human eyes, but very bright under the camera, so the facial recognition software cannot play its role. If the convenience and weirdness of such glasses are not considered, it seems to be effective. However, the high definition pixel of 2D camera nowadays can easily detect the eye information behind the glasses when such transparent glasses are under the visible light. Through algorithm, the facial information can be successfully collected.
Ophthalmologists and optics experts believe that the optical glass is the best medium to observe the world for the following reasons:
● Most stable performance in the complex service environment;
● High scratch resistance and mechanical stress;
● Glass materials with the minimum environmental impacts, 100% recycle;
The quality of our glasses starts from the glass quality control. Only the most innovative and high-performance material (namely, platinum glass) is used. It is a kind of high-quality optical glass, and its name comes from the production process. Some minerals called rare earth elements are added to enhance the best perception of color and contrast. Besides, through the platinum process, the refining is done to obtain the non-porous glass lenses. This is the process used by the well-known high-end glass lens brand in the world. We provide the same quality to you.
● Political officials are confirmed and killed by the hidden intelligent weapon station on the roadside using facial recognition.
● The private life of celebrities is made public as their face is recognized by the hidden camera.
● Under the monitoring recognition at the shopping mall, restaurant, hotel, school and concert, the personal privacy data are unethically dug and collected.
● When someone falls asleep, his mobile phone is unlocked by companions through facial recognition and his information is known at will.
● The monitoring camera on the road shoots the privacy behavior on the seat in the car and the identity is recognized and exposed on the internet.
●Someone is misjudged by the facial recognition system with high fault rate as the blacklist suspect and he is detained by the police.
● In the public places, the bare facial information is sold to the illegal institution after it is unethically captured.
● Someone legally participates in the anti-war parade, but he is wrongly made into the extremist list after his facial information is captured by the monitoring camera of the police.
The facial recognition AI of conventional access control system cannot operate when the user wears a mask and sunglasses. However, the access control system recognition is static and active cooperative recognition and the collected image has high quality. In this way, the smarter system algorithm can carry out the in-depth learning quickly, which improves the recognition rate. When we test the response of facial recognition system of access control system to the privacy glass for the first time, the green scanning box on the screen cannot find our facial position generally. However, when you wear the facial shelter and frequently conduct the test in front of the camera and fool the advanced system with smart algorithm, the system also carries out the in-depth learning through such opportunity, and can quickly master the comprehensive information of your face. Meanwhile,as for the community or company personnel management and check, the advanced system can confirm the identity based on body recognition technology (ReID). The body recognition technology is not completely reliant on facial recognition, but it carries out the identification based on dressing, shape, hair style and other characteristics. However, in the application scenario with extremely high security requirements, for example, financial payment scenario, algorithm will judge the results based on the mask shelter and lead the personnel whose facial shelter is serious to carry out the identity verification in other ways.
Therefore, it is suggested that the user who has worn the anti-facial recognition privacy glasses should not repeatedly and frequently test or fool the same recognition system. If the algorithm of such system is smart, the system may be more familiar with you compared with others when you challenge the system.
● Through nearly 30 years’ researches, under the controlled and even visible light conditions, the traditional facial recognition has developed greatly. At present, it can be widely used in military, finance, public security, daily life and other fields. However, the traditional facial recognition technologies include entirety method, manual design characteristic method and learning based characteristic description sub-method. The recognition accuracy rate slowly improves, but fails to achieve the ideal effects all the time. After the in-depth learning is applied on a large scale, the accuracy rate of facial recognition is greatly improved due to the introduction and development of in-depth learning. In 2014, the recognition accuracy rate of DeepFace in LFW data set exceeded human for the first time. With more researches and explorations subsequently, the accuracy rate of the latest in-depth learning facial recognition model has been more than 99.8% in LFW.
● Nowadays, many researches focus on the heterogeneous facial recognition under uncontrolled and non-visible light conditions. This is an outstanding issue. Heterogeneous face recognition refers to the facial recognition and matching across different visual domains. It not only involves the picture but also solves the facial image gap issue between different visual domains. Compared with the traditional facial recognition under visible light conditions, heterogeneous face recognition is filled with many severe challenges. In the real world, it has greater importance and value. For example, under the uncontrollable illumination conditions, the image should be shot by infrared ray at best. After that, it is matched with the visible light face in the database. Or, if it is impossible to accurately know the target face, a rough image is generated based on the known conditions at first. It is like the police drawing the sketch based on the descriptions of the eyewitness and then matching it with the real face.
● Due to the importance and value of heterogeneous face recognition and its severe challenges, it becomes more and more important and draws close attention in the facial recognition research. This domain is excessively dependent on the image shooting through infrared ray. The anti-facial recognition glasses pose a greater challenge to it. Interestingly, when the application of VCSEL laser chip technology of new facial detection technology started, the anti-facial recognition privacy glasses have been prepared to resist such new technology.
Privacy glasses are mainly used to prevent face recognition system equipment from grabbing face information accurately without authorization and matching with system database accurately, so as to reduce the harm of face recognition system to obtain identity and other privacy information. Ordinary sunglasses can only prevent the visible light monitoring of early monitoring devices. Now almost all devices have the perspective function of infrared light on sunglasses. Ordinary sunglasses can't successfully prevent the theft of face information, so they can't be called privacy glasses. On top of that, privacy.
In front of the early surveillance equipment, people wearing glasses posed challenges to face detection and face recognition software. Glasses, especially Sunglasses with reflective function, may hinder the general algorithm to find the required reference point in the face photos. Face detection and / or recognition products from different companies and products use different algorithms. Many of these early algorithms do recognize people. Wearing transparent glasses usually does not cause problems for face detection and recognition software, because key eye details are still visible. Even if they do cause some form of problem, the software will continue to examine the rest of the face with visible light. It is not uncommon that faces are still detected and recognized.
However, when people wear dark or shiny sunglasses, especially those with hidden eyes, the best algorithms in the world still have difficulties. When glasses cover or hide the distance between eyes, the problem becomes more complicated, and eyes are the key part of many algorithms. Most face recognition algorithms first identify eye features and compare them with pre-existing image databases before they enter other facial features. But all this was limited to the early single visible light device. Now, in the world, the combination of visible and infrared perspective monitoring equipment has been widely used.
Ordinary sunglasses in front of the monitoring device can not easily prevent the capture of face information. Except for special privacy sunglasses.
In the long run, there are two hidden dangers caused by the leakage of face data.
The first is face 3D data acquisition.
Now the popularity of 3D structured light in mobile phones is not high enough, but in the long run, the application of 3D structured light is becoming more and more widely, which has become an inevitable trend. At this time, the face recognition application will no longer obtain one or two simple photos, but may be financial level, security level 3D face data. If these data are leaked to lawless elements, it may really make 3D printing masks crack face recognition on a large scale. In particular, the unique biological information such as face will bring great trouble to individual users once it is copied.
The second is the association between face and other data tags.
At present, the popularity of face brushing payment is still relatively low, and there are few data tags associated with faces. But now, with the application of face brushing more and more widely, people need to be alert to associate face with various data tags without their knowledge. For example, some test loving applications are likely to launch some seemingly harmless small programs and combine them with face recognition. Unconsciously, the user's face information and interest tags may be packaged and sold to businesses. The next time you walk into a shopping mall, the salesperson may be able to see your preferences from the surveillance camera.
What kind of camera surveillance has face recognition function, the answer may be surprising! Which of the common hemispherical surveillance cameras, pan-tilt rotating cameras and pan-tilt outdoor spherical cameras on the market today are cameras for face recognition? The answer may be surprising -- they can all be used for face recognition with a simple back-end modification. Attempts to determine whether a camera is "face recognition" capable of how it looks are essentially impossible.
"In layman's terms, there are two ways to achieve face recognition. One is front-end recognition, which is currently the primary method of face recognition, where a chip with a face algorithm is installed in the camera; but in practice, there is also back-end recognition, which can be done by adding a back-end server for capturing faces, and back-end recognition methods are costly. But theoretically, all cameras, whether or not they have a face algorithm chip installed, can be modified through the back-end to achieve face recognition functionality.