Bibliographic citations
Mattus, J., (2021). Identificación de estados constructivos en viviendas en el Valle Chillón, empleando herramientas de aprendizaje automático (Machine Learning) [Universidad Nacional Agraria La Molina]. https://hdl.handle.net/20.500.12996/4655
Mattus, J., Identificación de estados constructivos en viviendas en el Valle Chillón, empleando herramientas de aprendizaje automático (Machine Learning) []. PE: Universidad Nacional Agraria La Molina; 2021. https://hdl.handle.net/20.500.12996/4655
@misc{renati/246447,
title = "Identificación de estados constructivos en viviendas en el Valle Chillón, empleando herramientas de aprendizaje automático (Machine Learning)",
author = "Mattus Zapata, Jose Carlos",
publisher = "Universidad Nacional Agraria La Molina",
year = "2021"
}
This research deepens in the application of machine learning trough the supervised classification in a house level scale. Here, we proposed the usage of two Machine Learning models (Covolutional Neural Networks and Naïve Bayes Classifier) for the identification of constructive states during the self-construction process. The examples are located in the Carabayllo district, Lima province. The research aims to adapt two algorithms of Machine Learning and compare their efficiency. To achieve this objective, we first gather free-usage images of the area of interest, conforming the basic data of the study. Following, we adapt the models codes for the computer vision task aiming the constructive states. In the first process, we select references points for the targeted blocks, using them we will extract freehigh resolution images. After that, the constructive states were classified manually, following validated methods in various researches. Resulting in a total of 117 georeferenced images. The second process involve the adaption of each models codes, taking into account the paths of the basic data, size of images, structure and parameters. Finally, a validation is performed using two different methods. Resulting in a “moderate” strength of agreement in the case of Naïve Bayes classifier, outperforming the Convolutional Neural Network in several bearings.
This item is licensed under a Creative Commons License