Özet:

Measures taken in areas such as tracking personnel, patients, students and criminals, protecting mobile devices and combating fraud have evolved with technological developments in artificial intelligence. Today, face recognition systems are used as one of the fast and precise solutions determined for this need, since the identification of the person and identity in these problems requires instantaneous and high accuracy. These systems are generally created by comparing the features in the face images taken from the picture, historical or live video with the features in the real image of the person previously taken. Face recognition systems can be integrated into many applications, as person and identity verification may be required in almost every sector. In this study, a face recognition system was developed in order to verify the driver using public transportation in the transportation sector. In the current system, drivers start the journey by operating the vehicles with their own personnel cards. However, the driver who is authorized to use the vehicle may violate the rules by handing his personal card to an unauthorized driver and risk the driving. For this reason, it has been understood that the personnel cards are insufficient for driver authorization. In order to prevent any accident and violation caused by unauthorized driving, it has become necessary to add a personnel recognition and identity verification module to the system. For this requirement, after the driver has verified his biometric data, it was decided that the verification should be repeated instantaneously, throughout the ride and at certain intervals so that the driver does not give the ride to another driver. By avoiding the methods such as fingerprint reader and iris verification that will distract the driver and risk the driving, a facial recognition system has been created to provide control with video images taken while driving through cameras that are currently on the vehicles and see the driver. In order to check the accuracy of the relevant system, a separate database was created for each driver with the images taken from the videos during the driving at different times. Based on pre-trained deep learning networks with pictures representing drivers, the system was tested with test images in databases using tensorflow and opencv libraries. Thus, it has been observed that the face recognition module developed can increase driving safety with authorized and verified personnel on the smart transportation system.

Anahtar Kelimeler: