Approaches to the protection of video surveillance systems when applying recognition algorithms

Asystemwith detection algorithms is a new step in thedevelopment of authentication methods. Currently these algorithms are used everywhere. Operating systems already have the facial recognition technology, as it allows a person to log in three times faster than when logging in with a password, and almost every smartphone today supports the function of unlocking with a camera. Object recognition, in turn, can beused for vehicleauthentication, when entering a controlled territory, and often plays an important role in implementing enterprise security.With obvious advantages, recognition systems are more likely to produce false results.Important characteristics of any biometric system are errors of the first and second kind.The paper identifies vulnerabilities in object and face recognition algorithms. Approaches to the protection of video surveillance systems when using recognition algorithms are defined. The analysis of the effectiveness of protective measures is carried out.Based on the results of theanalysis, it has been determinedthat additional identification methods cover all the vulnerabilities of the facial recognition algorithm, but this method is very expensive. Another effective measure of protection is the performance of specific actions by the user. This measure covers three of the six vulnerabilities, and is easy to implement and has a low cost.