Aplicación móvil para el reconocimiento automático de setas
Fecha
2021-12-09
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Jaén: Universidad de Jaén
Resumen
Este proyecto intenta integrar la tecnología en el ámbito micológico mediante el reconocimiento automático de
distintas especies de setas, diferenciando entre comestibles y tóxicas.
Gracias a la creciente disponibilidad de grandes volúmenes de imágenes, es posible realizar un estudio y
desarrollo de técnicas basadas en aprendizaje supervisado para la extracción de características de cada una de
las especies. Para conseguir esto, se han usado técnicas avanzadas de Deep Learning, más concretamente redes
neuronales convolucionales, que permiten reconocer patrones dentro de una imagen y poder extraer sus
características más relevantes.
Para facilitar el uso de esta red neuronal se desarrollará una aplicación móvil en la cuál se integrará la red
entrenada, sirviendo de ayuda al usuario para la catalogación de setas.
This project aims to provide novel tools in the mycology for the automatic recognition of different mushroom species, differentiating which are edible or toxic. Due to the increased availability of large image sets, in this study, we focus on techniques based on supervised learning for the extraction of key features of mushroom species. For this purpose, a Deep Learning method has been carried out, more specifically a convolutional neural network, which allows us to recognize image patterns and extract the most relevant characteristics for each one. In order to facilitate the use of the proposed method, a mobile application is developed in which the trained neural network and an easy-to-use interface are integrated, thus helping the user to catalog mushrooms.
This project aims to provide novel tools in the mycology for the automatic recognition of different mushroom species, differentiating which are edible or toxic. Due to the increased availability of large image sets, in this study, we focus on techniques based on supervised learning for the extraction of key features of mushroom species. For this purpose, a Deep Learning method has been carried out, more specifically a convolutional neural network, which allows us to recognize image patterns and extract the most relevant characteristics for each one. In order to facilitate the use of the proposed method, a mobile application is developed in which the trained neural network and an easy-to-use interface are integrated, thus helping the user to catalog mushrooms.