Byakugan: sistema de contagem autônoma de veículos em vias públicas

Abstract

This study is situated in the context of increasing urbanization and the need for efficient traffic management, proposing a software solution for automating vehicle counting on public roads. By applying big data, deep learning, and computer vision techniques, the aim is to provide a real-time solution useful for both governmental entities and private companies. The developed solution involves the implementation of a series of algorithms categorized into video pre-processing and vehicle detection and counting. The work also aims to create a data model to organize the information collected and a web page for data visualization. The adopted methodology includes the use of edge computing architecture for image pre-processing and a centralized server for heavier tasks, as well as the use of Kafka to manage data flow. As a result, the proposed solution is expected to contribute to the efficiency of traffic management and data-driven decision-making.