Bibliographic citations
Alarcon, J., Poma, R. (2021). Desarrollo de un algoritmo computacional de detección de equipos de protección eléctrica en personas, orientado a sistemas de vigilancia basados en cámaras IP [Tesis, Universidad Peruana de Ciencias Aplicadas (UPC)]. http://hdl.handle.net/10757/657930
Alarcon, J., Poma, R. Desarrollo de un algoritmo computacional de detección de equipos de protección eléctrica en personas, orientado a sistemas de vigilancia basados en cámaras IP [Tesis]. PE: Universidad Peruana de Ciencias Aplicadas (UPC); 2021. http://hdl.handle.net/10757/657930
@misc{renati/397646,
title = "Desarrollo de un algoritmo computacional de detección de equipos de protección eléctrica en personas, orientado a sistemas de vigilancia basados en cámaras IP",
author = "Poma Astete, Raul Felipe",
publisher = "Universidad Peruana de Ciencias Aplicadas (UPC)",
year = "2021"
}
This project proposes the implementation of a computational algorithm for the detection of personal protective equipment (PPE) oriented to surveillance systems based on IP cameras. The proposed algorithm carried out the autonomous monitoring of head and hand PPE, reducing occupational risks and taking care of the health of the workers who carry out the activities. The proposed system uses artificial vision algorithms to process the images captured by the company's video surveillance cameras. In a first stage, the Open Pose neural network is used to extract the images of the workers' arms and heads. The images are then processed by convolutional neural networks trained to classify the existence of the mentioned EPPs. The system will be applied to the reality of the company Distribución Eléctrica S.A, where a software will be implemented that allows end users to view the alerts generated by the algorithm, as well as receive the alerts through their email. The results obtained by the classification system will be evaluated using metrics such as precision, sensitivity, and specificity.
This item is licensed under a Creative Commons License