Publication: Detección de vehículos en movimiento en vídeos mediante técnicas de aprendizaje profundo
Loading...
Official URL
Full text at PDC
Publication Date
2021
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
En este trabajo se presenta una aplicación para la identificación de vehículos en vías urbanas mediante técnicas de detección del movimiento en imágenes y aprendizaje profundo para su identificación.
Mediante el entrenamiento de redes neuronales convolucionales como AlexNet y GoogLeNet se procede a la detección de vehículos en una serie de videos previamente cargados por el usuario.
La aplicación permite al usuario ajustar los parámetros de entrenamiento de las redes neuronales, realizar dicho entrenamiento y cargar un video para la posterior detección de los vehículos. Adicionalmente, se procede a la carga de dichos resultados a la red social Twitter para su difusión.
En el trabajo se realiza un análisis y valoración de resultados, tanto en la fase de entrenamiento como en la de decisión.
Identification of vehicles on urban roads using motion detection techniques in images and deep learning for their identification. By training convolutional neural networks such as AlexNet or GoogLeNet, vehicles are detected in a series of videos previously uploaded by the user. The present work will consist on the creation of an application in which the user is allowed to adjust the training parameters of the neural networks, carry out said training and upload a video for the subsequent detection of the vehicles. Additionally, these results are uploaded to the social network Twitter for dissemination. In this work, an analysis and evaluation of results is carried out, both in the training phase and in the decision phase.
Identification of vehicles on urban roads using motion detection techniques in images and deep learning for their identification. By training convolutional neural networks such as AlexNet or GoogLeNet, vehicles are detected in a series of videos previously uploaded by the user. The present work will consist on the creation of an application in which the user is allowed to adjust the training parameters of the neural networks, carry out said training and upload a video for the subsequent detection of the vehicles. Additionally, these results are uploaded to the social network Twitter for dissemination. In this work, an analysis and evaluation of results is carried out, both in the training phase and in the decision phase.
Description
Trabajo de Fin de Grado en Ingeniería del Software, Facultad de Informática UCM, Departamento de Ingeniería del Software e Inteligencia Artificial, Curso 2020/2021, descargar el código del repositorio de Github: https://github.com/JesusGranizo/Codigo