Bibliographic citations
Vera, K., Chala, D. (2023). Arquitectura de red neuronal convolucional para la identificación de especies de árboles que evite la interrupción del flujo eléctrico [Tesis, Universidad Peruana de Ciencias Aplicadas (UPC)]. http://hdl.handle.net/10757/667769
Vera, K., Chala, D. Arquitectura de red neuronal convolucional para la identificación de especies de árboles que evite la interrupción del flujo eléctrico [Tesis]. PE: Universidad Peruana de Ciencias Aplicadas (UPC); 2023. http://hdl.handle.net/10757/667769
@misc{renati/404994,
title = "Arquitectura de red neuronal convolucional para la identificación de especies de árboles que evite la interrupción del flujo eléctrico",
author = "Chala Gonzalez, Diego David",
publisher = "Universidad Peruana de Ciencias Aplicadas (UPC)",
year = "2023"
}
The purpose of this research work is to propose the identification of tree species using images through convolutional neural networks for cadastral processes in electric power companies. With this proposal, it is intended to offer these companies a tool that allows them to improve the identification of trees when pruning is needed and thus be able to reduce the margin of error compared to traditional measure methods to avoid power outages due to tree growth. To develop this project, an investigation was carried out on the possible alternatives that would help to identify the plant species. An analysis of machine learning tools was carried out, which resulted in the convolutional neural network being the best option to be able to identify, segment and discriminate with image capture. In addition, the characteristics of the most common plant species were monitored, and it was decided to use a dataset to manage these important characteristics by discriminating between plant species and using them together with neural networks for their identification. Next, the convolutional neural network architecture for plant identification was designed. Finally, a survey was carried out on the level of user satisfaction on the advantages of the implementation of the convolutional neural network, such as the reduction of plant identification time, the easy use, and the total elimination of operating errors, in said survey; the boss and users of the process participated. With the results of said survey, it was possible to validate that the degree of user satisfaction is high with respect to the advantages offered by the implementation of the proposed architecture; which meets the objectives of being able to identify the species of tree and thus not make mistakes in pruning and interrupt electrical flows.
This item is licensed under a Creative Commons License