Thesis Plataforma IoT para detección de aglomeraciones en espacios públicos con énfasis en Metro de Santiago
No Thumbnail Available
Date
2026-10
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Telemática
Departament
Campus
Campus Santiago San Joaquín
Abstract
Este trabajo aborda la problemática asociada a las aglomeraciones de personas en espacios públicos de alta demanda, donde las variaciones en la afluencia dificultan la supervisión y gestión oportuna de los flujos. Los mecanismos tradicionales de monitoreo, basados principalmente en la observación de cámaras de videovigilancia, pueden presentar limitaciones para obtener de manera automática información sobre la cantidad y distribución de personas presentes en un espacio. En este contexto, el propósito de esta memoria es desarrollar e implementar un sistema de monitoreo inteligente de afluencia de personas que permita generar información para apoyar la toma de decisiones operativas. Para su desarrollo se considera Metro de Santiago como caso de estudio. La solución implementada integra distintas etapas asociadas a la captura, procesamiento, almacenamiento y visualización de información proveniente de fuentes de video. El procesamiento se realiza mediante técnicas de visión por computador e inteligencia artificial, utilizando YOLOv8 para la detección de personas y la obtención de información espacial mediante segmentación. A partir de estos resultados, el sistema genera dos indicadores principales: el conteo de personas y un índice de aglomeración que representa la proporción del área ocupada por las personas detectadas. La información obtenida se almacena en una base de datos MySQL y es gestionada mediante un backend desarrollado con FastAPI. Además, se implementa una interfaz web mediante React que permite visualizar las fuentes de video procesadas, consultar la evolución temporal de los indicadores, gestionar cámaras y alertas, y generar reportes históricos en formato CSV. Para evaluar el comportamiento del sistema se utilizan grabaciones correspondientes al caso de estudio y a una fuente de prueba local, considerando distintos escenarios de presencia y distribución de personas. Las pruebas permiten analizar (...).
This work addresses the problem of crowding in high-demand public spaces, where fluctuations in pedestrian flow make timely supervision and management difficult. Traditional monitoring methods, which rely mainly on operators watching video surveillance cameras, have limited ability to automatically obtain information on the number and distribution of people in a given space. In this context, this thesis aims to develop and implement an intelligent pedestrian flow monitoring system that generates information to support operational decision-making, using the Metro from Santiago, Chile, as a case study. The solution integrates the capture, processing, storage, and visualization of information from video sources. Processing relies on computer vision and artificial intelligence techniques, using YOLOv8 for people detection and segmentation to extract spatial information. From these results, the system produces two main indicators: the people count and a crowding index, defined as the proportion of the area occupied by detected people. The data is stored in a MySQL database and managed through a backend built with FastAPI. In addition, a web interface developed with React allows users to view the processed video sources, examine how the indicators evolve over time, manage cameras and alerts, and generate historical reports in CSV format. To evaluate the system, recordings from the case study and a local test source are used, covering different scenarios in terms of the presence and distribution of people. The tests analyze (...).
This work addresses the problem of crowding in high-demand public spaces, where fluctuations in pedestrian flow make timely supervision and management difficult. Traditional monitoring methods, which rely mainly on operators watching video surveillance cameras, have limited ability to automatically obtain information on the number and distribution of people in a given space. In this context, this thesis aims to develop and implement an intelligent pedestrian flow monitoring system that generates information to support operational decision-making, using the Metro from Santiago, Chile, as a case study. The solution integrates the capture, processing, storage, and visualization of information from video sources. Processing relies on computer vision and artificial intelligence techniques, using YOLOv8 for people detection and segmentation to extract spatial information. From these results, the system produces two main indicators: the people count and a crowding index, defined as the proportion of the area occupied by detected people. The data is stored in a MySQL database and managed through a backend built with FastAPI. In addition, a web interface developed with React allows users to view the processed video sources, examine how the indicators evolve over time, manage cameras and alerts, and generate historical reports in CSV format. To evaluate the system, recordings from the case study and a local test source are used, covering different scenarios in terms of the presence and distribution of people. The tests analyze (...).
Description
Keywords
Monitoreo de aglomeraciones, YOLOv8, Visión por computador, Detección de personas, Índice de aglomeración, Afluencia de personas, Inteligencia artificial
