Diseño e implementación de una solución tecnológica para incrementar la adopción de chalecos en trabajadores de obras civiles y almacenes
Cargando...
Fecha
2023
Autores
Cujia Ramirez, Katherine Daniela
De Arco Escorcia, Joigmar Jesus
Farrayans Romero, Sebastian
Yance Orozco, Ariel Armel
Título de la revista
ISSN de la revista
Título del volumen
Editor
Ediciones Universidad Simón Bolívar
Facultad de Ingenierías
Facultad de Ingenierías
Resumen
Los equipos de protección personal (PPE) son requeridos en las industrias dentro de sus espacios de trabajo, como lo estipula el reglamento interno de la empresa, el cual establece las normativas sobre lo que se puede o no hacer dentro de las instalaciones. En particular, los chalecos reflectivos desempeñan una función esencial al asegurar la visibilidad del usuario. Su principal objetivo es permitir que cualquier persona que se encuentre en las proximidades del área pueda identificar al usuario desde una distancia considerable. Este proyecto tiene como objetivo desarrollar un sistema de detección y monitoreo del uso de chalecos reflectivos en entornos laborales, utilizando cámaras ubicadas dentro de los lugares de trabajo, bodegas y plantas. En tiempo real, el sistema es capaz de identificar si las personas llevan puestos o no los chalecos reflectivos mediante un proceso que involucra el procesamiento de imágenes con OpenCV, Numpy y MediaPipe. Luego, se lleva a cabo el entrenamiento de redes neuronales utilizando TensorFlow. Posteriormente, se vuelve a utilizar MediaPipe para detectar a las personas y extraer los píxeles donde se encuentran, lo cual mejora la eficiencia del entrenamiento de las redes neuronales y genera una respuesta precisa para determinar si una persona lleva o no puesto un chaleco reflectivo.
Finalmente, este método logra una precisión del 73%. Si bien el sistema de detección funciona adecuadamente en entornos con buena iluminación, aún no ha sido probado en entornos reales, como almacenes.
Personal Protective Equipment (PPE) is required in industries within their workspaces, as stipulated by the company's internal regulations, which establish guidelines on what can and cannot be done within the facilities. In particular, reflective vests play an essential role in ensuring user visibility. Their main objective is to allow anyone in the vicinity to identify the user from a considerable distance. This project aims to develop a system for detecting and monitoring the use of reflective vests in workplace environments, using cameras located within workplaces, warehouses, and plants. In real-time, the system can identify whether individuals are wearing reflective vests or not through a process involving image processing with OpenCV, Numpy, and MediaPipe. Neural network training is then carried out using TensorFlow. Subsequently, MediaPipe is used again to detect individuals and extract the pixels where they are located, improving the efficiency of neural network training and providing an accurate response to determine whether a person is wearing a reflective vest or not. Finally, this method achieves an accuracy of 73%. Although the detection system works well in environments with good lighting, it has not yet been tested in real environments such as warehouses.
Personal Protective Equipment (PPE) is required in industries within their workspaces, as stipulated by the company's internal regulations, which establish guidelines on what can and cannot be done within the facilities. In particular, reflective vests play an essential role in ensuring user visibility. Their main objective is to allow anyone in the vicinity to identify the user from a considerable distance. This project aims to develop a system for detecting and monitoring the use of reflective vests in workplace environments, using cameras located within workplaces, warehouses, and plants. In real-time, the system can identify whether individuals are wearing reflective vests or not through a process involving image processing with OpenCV, Numpy, and MediaPipe. Neural network training is then carried out using TensorFlow. Subsequently, MediaPipe is used again to detect individuals and extract the pixels where they are located, improving the efficiency of neural network training and providing an accurate response to determine whether a person is wearing a reflective vest or not. Finally, this method achieves an accuracy of 73%. Although the detection system works well in environments with good lighting, it has not yet been tested in real environments such as warehouses.
Descripción
Palabras clave
OpenCV, Procesamiento de Imágenes, MediaPipe, Chalecos reflectivos, Seguridad, Almacenes, Image Processing, MediaPipe, Reflective vests, Security, Warehouses