Lectura de medidores de energía por medio de sistemas de reconocimiento de caracteres con Machine learning - Redes neuronales (OCR)

Public service companies in the context of Latin America make their consumption reading for billing through collectors who are responsible for taking photographs and manually writing the consumption, then they require more personnel to verify the images taken in the field work, which which makes the billing process time consuming, inefficient and costly. The present research developed a solution for meter reading and thus optimize and automate these processes by means of low-cost technological tools with the use of artificial intelligence. In this sense, an application with the ability to quickly read power consumption devices was developed by means of deep learning techniques, specifically convolutional neural networks with the YOLO-V4 architecture. For this, in technical terms the three processes -specific objectives- the data was prepared for the training of the models, the training data was increased for better performance...

• Autores: Salazar Ortiz, Iván Arturo
• Tipo de recurso: Trabajo de grado de pregrado
• Fecha de publicación: 2020
• Institución: Universidad de los Andes
• Repositorio: Séneca: repositorio Uniandes
• Idioma: spa
• OAI Identifier: oai:repositorio.uniandes.edu.co:1992/51502
• Acceso en línea: http://hdl.handle.net/1992/51502
• Palabra clave: Contadores eléctricos; Electricidad; Consumo de energía eléctrica; Sistemas de reconocimiento de configuraciones; Aprendizaje automático (Inteligencia artificial); Redes neuronales convolucionales; Ingeniería
• Rights: openAccess
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/