Diseño y desarrollo de aplicación móvil para la clasificación de flora nativa chilena utilizando redes neuronales convolucionales
Fecha
2022-01-22
Profe guía
Perfil ORCID
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidade Federal do Paraná
ISBN
ISSN
2237-826X
ISSNe
Resumen
Introducción: Las aplicaciones móviles, a través de la visión artificial, son capaces de reconocer especies vegetales en tiempo real. Sin embargo, las actuales aplicaciones de reconocimiento de especies no consideran la gran variedad de especies endémicas y nativas de Chile, tendiendo a predecir erróneamente. Esta investigación presenta la construcción de un dataset de especies chilenas y el desarrollo de un modelo de clasificación optimizado e implementado en una aplicación móvil. Método: La construcción del dataset se realizó a través de la captura de fotografías de especies en terreno y selección de imágenes de datasets en línea. Se utilizaron redes neuronales convolucionales para desarrollar los modelos de predicción de imágenes. Se realizó un análisis de sensibilidad al entrenar las redes, validando con k-fold cross validation y efectuando pruebas con distintos hiperparámetros, optimizadores, capas convolucionales y tasas de aprendizaje, para seleccionar los mejores modelos y luego ensamblarlos en un solo modelo de clasificación. Resultados: El dataset construido se conformó por 46 especies, incluyendo especies nativas, endémicas y exóticas de Chile, con 6120 imágenes de entrenamiento y 655 de prueba. Los mejores modelos se implementaron en una aplicación móvil, donde se obtuvo un porcentaje de acierto de aproximadamente 95% con respecto al conjunto de pruebas. Conclusiones: La aplicación desarrollada es capaz de clasificar especies correctamente con una probabilidad de acierto acorde con el estado del arte de la visión artificial y de mostrar información de la especie clasificada.
Introduction: Mobile apps, through artificial vision, are capable of recognizing vegetable species in real time. However, the existing species recognition apps do not take in consideration the wide variety of endemic and native (Chilean) species, which leads to wrong species predictions. This study introduces the development of a chilean species dataset and an optimized classification model implemented to a mobile app. Method: the data set was built by putting together pictures of several species captured on the field and by selecting some pictures available from other datasets available online. Convolutional neural networks were used in order to develop the images prediction models. The networks were trained by performing a sensitivity analysis, validating with k-fold cross validation and performing tests with different hyper-parameters, optimizers, convolutional layers, and learning rates in order to identify and choose the best models and then put them together in one classification model. Results: The final data set was compounded by 46 species, including native species, endemic and exotic from Chile, with 6120 training pictures and 655 testing pictures. The best models were implemented on a mobile app, obtaining a 95% correct prediction rate with respect to the set of tests. Conclusion: The app developed in this study is capable of classifying species with a high level of accuracy, depending on the state of the art of the artificial vision and it can also show relevant information related to the classified species.
Introduction: Mobile apps, through artificial vision, are capable of recognizing vegetable species in real time. However, the existing species recognition apps do not take in consideration the wide variety of endemic and native (Chilean) species, which leads to wrong species predictions. This study introduces the development of a chilean species dataset and an optimized classification model implemented to a mobile app. Method: the data set was built by putting together pictures of several species captured on the field and by selecting some pictures available from other datasets available online. Convolutional neural networks were used in order to develop the images prediction models. The networks were trained by performing a sensitivity analysis, validating with k-fold cross validation and performing tests with different hyper-parameters, optimizers, convolutional layers, and learning rates in order to identify and choose the best models and then put them together in one classification model. Results: The final data set was compounded by 46 species, including native species, endemic and exotic from Chile, with 6120 training pictures and 655 testing pictures. The best models were implemented on a mobile app, obtaining a 95% correct prediction rate with respect to the set of tests. Conclusion: The app developed in this study is capable of classifying species with a high level of accuracy, depending on the state of the art of the artificial vision and it can also show relevant information related to the classified species.
Descripción
Lugar de Publicación
Sponsorship
Citación
AtoZ: Novas Práticas em Informação e Conhecimento, Vol 11 N° 1 (Año 2022) p 1-13.
Palabras clave
Licencia
Atribución-NoComercial-CompartirIgual 3.0 Chile (CC BY-NC-SA 3.0 CL)