- Non-Linear Thinking is human thought characterized by expansion in multiple directions, rather than in one direction, and based on the concept that there are multiple starting points from which one can apply logic to a problem.
- There is a difference between the Viola-Jones detection framework and how the masks are used in the neural networks.
- Grigory Bakunov and his group from Russia have developed an application that generates variants of makeup that interfere with identification systems.
Facial recognition technology has become one of the most utilized tools in airports and other entities. It is a great tool for risk assessment and controlling ational security around the globe. Facial recognition projects have been around since last century. There were many primitive attempts that took place in the 1980s.
The most notable project, and a predecessor of the currently evolved facial recognition technology, was development by Paul Viola and Michael Jones in 2001. The Viola–Jones object detectionframework is the first framework to provide competitive object detection rates in real-time. At the time a standard available computer could produce 15 images per second using this method.
The convolutional neural network is “sharpened” to extract the most characteristic features from an image, and to do this automatically and at different levels of abstraction even on the smallest photo. While the first levels usually respond to simple patterns like gradient, clear borders, etc., the complexity of features increases with each new level. Masks that the neural network tries on at high levels often actually resemble human faces or fragments of them. In addition, unlike the principal component method, neural networks combine features in a non-linear way.
Non-Linear Thinking is human thought characterized by expansion in multiple directions, rather than in one direction, and based on the concept that there are multiple starting points from which one can apply logic to a problem.
There is a difference between the Viola-Jones detection framework and how the masks are used in the neural networks which are a set of algorithms, modeled loosely after the human brain, that are designed to recognize patterns. They interpret sensory data through a kind of machine perception, labeling or clustering raw input. Such systems “learn” to perform tasks by considering examples. Therefore, with human aid the masks are the learning process which includes the utilization of a myriad of facial images and diverse backgrounds.
There after a triples method is formed. The predecessor is Pythagorean triples that have been around for centuries, just not utilized in the digital world, since at the time, such didn’t exist. For the purposes of the neural networks the triples mean a set of images in which the first two pictures are of the same face of the same person. The third image is of a completely different person. The neural network is trained to identify the features to realize the two images are of the same person, based on identifying their specific features.
Russia has been very busy working and perfecting this technology. This is not a surprise considering Putin’s desire to have control of the whole nation. Russian deepvo V3 algorithm of VOCORD achieves a 92% result. There is a a caveat to this technology: it includes patents purchased by a Chinese Huawei technology giant.
Huawei Technologies is a multinational technology company that provides telecommunications equipment and sells consumer electronics, smartphones and is headquartered in Shenzhen, Guangdong. This is why there is a great concern pertaining to the use of Huawei technologies in the West. This technology can be used to obtain facial recognition data of western nationals and to be used for the benefit of Chinese intelligence.
The only way to “trick” the facial recognition system is to use certain colors and patterns. For example, Grigory Bakunov and his group from Russia have developed an application that generates variants of makeup that interfere with the identification systems. However, the system is not available in open source, but utilized by Russian intelligence services to aid in disguise of their agents traveling around the world and to confuse airport facial recognition systems. There are companies that make certain types of pattern for clothing that can possibly help in diversion.