Publication: Algoritmo de pre-procesamiento de imágenes para la estimación de la edad
Loading...
Official URL
Full text at PDC
Publication Date
2021-06
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Hoy en día, el campo de las redes neuronales se encuentra en plena expansión y son muchos los avances que se realizan. Los sistemas de reconocimiento facial han ganado un gran peso en la sociedad en los ´últimos años, incorporándose en el día a día de las personas, por ejemplo, para desbloquear el teléfono móvil. Sin embargo, existen otras muchas tareas por realizar, siendo una de las más importantes el reconocimiento de la edad a través de una imagen. El principal problema encontrado para realizar esta tarea es la poca cantidad de bases de datos dedicadas al tema y la poca calidad de las imágenes en las existentes. En este trabajo se intenta crear diferentes modelos y realizar un pre-procesamiento de las imágenes, para que el entrenamiento de los modelos se realice de manera más precisa. Para cumplir este objetivo, se ha investigado acerca de los conceptos relativos al campo de la estimación de la edad en los trabajos previos relacionados con el tema y en las técnicas más utilizadas y con mejores resultados a la hora de preprocesar imágenes. Finalmente se han probado los conocimientos adquiridos mediante diferentes experimentos para medir la influencia del pre-procesamiento de imágenes en los diferentes modelos creados.
Nowadays, neural networks are currently being used to develop many intelligent applications. Facial recognition systems have gained great importance in society in recent years, being incorporated into people’s daily lives, for example, to unlock cell phones. However, there are many other tasks to be solved, one of the most important ones is the age recognition using an image. The more important issue encountered to carry out this task is the small number of databases dedicated to the subject and the poor quality of the images in the existing ones.w Therefore, in this work we try to create different models and implement a pre-processing of the images, so that the training of the models is performed more accurately. To achieve this goal, extensive research has been carried out on the concepts encompassing the field of age estimation, on previous work related to the subject as well as the most widely used and best performing techniques used to preprocess images. Finally, the knowledge acquired has been tested by means of different experiments to measure the influence of image pre-processing on the different models created.
Nowadays, neural networks are currently being used to develop many intelligent applications. Facial recognition systems have gained great importance in society in recent years, being incorporated into people’s daily lives, for example, to unlock cell phones. However, there are many other tasks to be solved, one of the most important ones is the age recognition using an image. The more important issue encountered to carry out this task is the small number of databases dedicated to the subject and the poor quality of the images in the existing ones.w Therefore, in this work we try to create different models and implement a pre-processing of the images, so that the training of the models is performed more accurately. To achieve this goal, extensive research has been carried out on the concepts encompassing the field of age estimation, on previous work related to the subject as well as the most widely used and best performing techniques used to preprocess images. Finally, the knowledge acquired has been tested by means of different experiments to measure the influence of image pre-processing on the different models created.
Description
Trabajo de Fin de Doble Grado en Ingeniería Informática y Matemáticas, Facultad de Informática UCM, Departamento de Ingeniería del Software e Inteligencia Artificial, Curso 2020/2021.
El trabajo se puede encontrar en el siguiente enlace: https://gitlab.fdi.ucm.es/miguel.franqueira/ ageestimation