Proposal for a system of recognition of Salvadoran plaques for the University of El Salvador

Abstract

Introduction: This paper provides a proposal for a Salvadoran license plate recognition system for the University of El Salvador. To arrive at this proposal, candidates based on free software were sought in the most popular software repositories on the Internet. Objective: to propose a Salvadoran license plate recognition system for the University of El Salvador. Method: experimental, which allowed obtaining data for processing and the results could be comparable, candidates based on free software were searched for in the most popular software repositories on the Internet. The selected candidates were OpenALPR and the combination of YOLOv5 with PaddleOCR, and they were installed in a test environment. Then a set of images of Salvadoran plates was created to train and compare the models. The inferences of each candidate were obtained for each image in the set, as well as the hits, time used and certainty. From this information, the candidates were compared with each other. Result: in each element of comparison, the candidate with the best performance was the combination of YOLOv5 with PaddleOCR, becoming the proposed recognition system for Salvadoran license plates for the University of El Salvador.