Automatic Number Plate Recognition Technologies for Azerbaijani Vehicle Registration Plates
Date Issued
2022-04
Author(s)
Huseynov, Kamran
Abstract
Vehicle monitoring on road surveillance and parking lot management is one of the global
problems in the computer vision field that requires the application of practical and
contemporary technologies. The main applications in the computer vision field concerning
these problems are called Automatic Number Plate Recognition Systems which are based on
ever changing methodologies given the rapid development in Artificial Intelligence and Deep
Learning fields. The reason for the broad spectrum of different methodologies is that these
systems are in high demand by institutions and private businesses where ANPR systems are
frequently upgraded to match the best practices available in order to improve performance,
increase speed, reduce cost, and overall maximize the efficiency. Thus, this paper is focused on
current practices in Automatic Number Plate Recognition systems and the performance of
individual building blocks in the ANPR pipeline. In this paper we will examine in depth the
details of contemporary practices in the computer vision field for ANPR solutions, especially,
the solutions considering the recent innovations and enhancements in deep neural networks.
Even though vehicle monitoring is a global problem, We will specifically focus on vehicle
registration numbers issued by Azerbaijan and challenges regarding this selected domain.
Given the domain, this paper will finally draw comprehensive comparisons between current
ANPR technologies and derive meaningful conclusions based on data, experiments, and results
problems in the computer vision field that requires the application of practical and
contemporary technologies. The main applications in the computer vision field concerning
these problems are called Automatic Number Plate Recognition Systems which are based on
ever changing methodologies given the rapid development in Artificial Intelligence and Deep
Learning fields. The reason for the broad spectrum of different methodologies is that these
systems are in high demand by institutions and private businesses where ANPR systems are
frequently upgraded to match the best practices available in order to improve performance,
increase speed, reduce cost, and overall maximize the efficiency. Thus, this paper is focused on
current practices in Automatic Number Plate Recognition systems and the performance of
individual building blocks in the ANPR pipeline. In this paper we will examine in depth the
details of contemporary practices in the computer vision field for ANPR solutions, especially,
the solutions considering the recent innovations and enhancements in deep neural networks.
Even though vehicle monitoring is a global problem, We will specifically focus on vehicle
registration numbers issued by Azerbaijan and challenges regarding this selected domain.
Given the domain, this paper will finally draw comprehensive comparisons between current
ANPR technologies and derive meaningful conclusions based on data, experiments, and results