RECONHECIMENTO ÓPTICO DE PLACAS DE VEÍCULOS: UMA ABORDAGEM PARA APRIMORAR O GERENCIAMENTO DE ENTRADAS E SAÍDAS EM PÁTIOS
Carregando...
Data
2023-12
Autores
Orientador
Pretto, Fabrício
Banca
Wolf, Alexandre Sturmer
Diemer, Mouriac Halen
Título do periódico
ISSN
Título do Volume
Editor
Resumo
O gerenciamento de veículos envolve a organização e controle eficiente de frotas, visando otimizar a utilização dos veículos, garantir a segurança, reduzir custos e melhorar a eficiência operacional. Dentro do contexto industrial, que abrange uma variedade de setores, esse gerenciamento estende-se à organização e controle de acesso de veículos, visando assegurar a agilidade, segurança e eficiência na gestão de pátio. No entanto, em muitas indústrias, esse processo de gerenciamento é realizado de forma manual, causando a demora e ineficiência do procedimento, levando a erros de registro, longos períodos de espera para os motoristas e sobrecarga nas portarias. Levando-se em consideração estes aspectos, propôs-se um sistema de reconhecimento óptico de placas de veículos para validação automatizada de acesso. Esse sistema pretende otimizar fluxos, garantir a segurança e melhorar a eficiência do processo. Para tanto, construiu-se uma arquitetura de projeto, que abrange as áreas de conhecimento de processamento de imagens, reconhecimento óptico de caracteres (OCR), Deep Learning e controle de acesso via software. Tanto para o reconhecimento das placas, quanto para a leitura dos caracteres presentes nestas, utilizou-se tecnologias como TypeScript, Python, algoritmo YOLO, OCR e Google Cloud Platform. No que diz respeito ao controle de acesso, adotou-se tecnologias como TypeScript, MySQL e aproveitou-se os resultados obtidos pelo reconhecimento prévio das placas. Nesse sentido, a implementação do projeto possibilitou o reconhecimento automático de veículos por meio do processamento de imagens e análise de atributos, como placas de identificação e autorizações de entrada. Ao final, o experimento foi validado com 287 veículos distintos estacionados em empresas de pequeno e médio porte de diversos segmentos, sendo que, uma das principais empresas utilizadas, é do segmento de nutrição animal, que atualmente realiza o processo de gestão de pátio e entrada de veículos de forma manual. Dessa forma, foi possível validar que a solução atingiu um processamento eficiente do reconhecimento de placas e do gerenciamento de acesso em um tempo de processamento consideravelmente baixo, onde obteve-se um tempo médio de processamento total de 11,1 segundos. Outro ponto importante, refere-se a assertividade, a qual atingiu 91,85%, e mostrou-se aceitável, contribuindo, também, para a minimização de erros de registro e proporcionando, consequentemente, um aumento na segurança e confiabilidade dos dados.
Vehicle management involves the efficient organization and control of fleets, aiming to optimize vehicle usage, ensure safety, reduce costs, and enhance operational efficiency. Within the industrial context, which spans a variety of sectors, this management extends to the organization and control of vehicle access, aiming to ensure agility, security, and efficiency in yard management. However, in many industries, this management process is conducted manually, resulting in delays and inefficiencies, leading to recording errors, long waiting periods for drivers, and congestion at entry points. Taking these aspects into consideration, a system of optical vehicle plate recognition for automated access validation was proposed. This system aims to optimize flows, ensure security, and improve process efficiency. To achieve this, a project architecture was built, covering areas of knowledge such as image processing, optical character recognition (OCR), Deep Learning, and software-based access control. For both plate recognition and reading of characters, technologies such as TypeScript, Python, YOLO algorithm, OCR, and Google Cloud Platform were used. Regarding access control, technologies such as TypeScript and MySQL were adopted, leveraging the results obtained from the previous plate recognition. In this sense, the project implementation enabled automatic recognition of vehicles through image processing and analysis of attributes, such as identification plates and entry authorizations. In the end, the experiment was validated with 287 different vehicles parked in small and medium-sized companies from various sectors. One of the main companies operates in the animal nutrition sector, currently managing yard and vehicle entry manually. Thus, it was possible to validate that the solution achieved efficient processing of plate recognition and access management in a considerably short processing time, with an average total processing time of 11.1 seconds. Another important point is the accuracy, which reached 91.85% and proved to be acceptable, contributing to the minimization of registration errors and consequently increasing the security and reliability of the data.
Vehicle management involves the efficient organization and control of fleets, aiming to optimize vehicle usage, ensure safety, reduce costs, and enhance operational efficiency. Within the industrial context, which spans a variety of sectors, this management extends to the organization and control of vehicle access, aiming to ensure agility, security, and efficiency in yard management. However, in many industries, this management process is conducted manually, resulting in delays and inefficiencies, leading to recording errors, long waiting periods for drivers, and congestion at entry points. Taking these aspects into consideration, a system of optical vehicle plate recognition for automated access validation was proposed. This system aims to optimize flows, ensure security, and improve process efficiency. To achieve this, a project architecture was built, covering areas of knowledge such as image processing, optical character recognition (OCR), Deep Learning, and software-based access control. For both plate recognition and reading of characters, technologies such as TypeScript, Python, YOLO algorithm, OCR, and Google Cloud Platform were used. Regarding access control, technologies such as TypeScript and MySQL were adopted, leveraging the results obtained from the previous plate recognition. In this sense, the project implementation enabled automatic recognition of vehicles through image processing and analysis of attributes, such as identification plates and entry authorizations. In the end, the experiment was validated with 287 different vehicles parked in small and medium-sized companies from various sectors. One of the main companies operates in the animal nutrition sector, currently managing yard and vehicle entry manually. Thus, it was possible to validate that the solution achieved efficient processing of plate recognition and access management in a considerably short processing time, with an average total processing time of 11.1 seconds. Another important point is the accuracy, which reached 91.85% and proved to be acceptable, contributing to the minimization of registration errors and consequently increasing the security and reliability of the data.
Descrição
Palavras-chave
gerenciamento de veículos; reconhecimento óptico de placas; processamento de imagens; vehicle management; optical plate recognition; image processing
Citação
SCHOENINGER, Fáiza Letícia. RECONHECIMENTO ÓPTICO DE PLACAS DE VEÍCULOS: UMA ABORDAGEM PARA APRIMORAR O GERENCIAMENTO DE ENTRADAS E SAÍDAS EM PÁTIOS. 2023. Monografia (Graduação em Engenharia de Software) – Universidade do Vale do Taquari - Univates, Lajeado, 13 dez. 2023. Disponível em: http://hdl.handle.net/10737/4304.