Análisis de los Átomos del Contenedor Multimedia de Vídeos
Loading...
Official URL
Full text at PDC
Publication date
2023
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citation
Abstract
En la actualidad, la sociedad tiende a capturar y compartir vídeos gracias al empleo de redes sociales o simplemente el hecho de contar con un dispositivo con cámara o software para editar dicho archivo multimedia. Esto hace que sea difícil el poder detectar la manipulación que se ha llevado a cabo en el archivo. En este trabajo se propone un análisis de los contenedores multimedia para comprobar si sus resultados son equiparables a los de un software de detección de manipulación. Se extraen los átomos que son los datos que conforman el contenedor, para tener constancia del rastro que dejan los distintos programas y después realizar una serie de experimentos con conjuntos de datos diferentes
para la obtención del mejor modelo. Los resultados demuestran que este tipo de análisis tiene resultados excelentes con algoritmos de aprendizaje automático, lo que hace que sean modelos muy robustos y explicables a la hora de detectar los átomos clave.
Nowadays, society tends to capture and share videos thanks to the use of social networks or simply the fact of having a device with a camera or software to edit said multimedia file. This makes it difficult to detect the manipulation that has taken place in the file. This work proposes an analysis of multimedia containers to check if their results are comparable to those of a manipulation detection software. The atoms that are the data that make up the container are extracted, to be aware of the trace left by the different programs and then perform a series of experiments with different data sets to obtain the best model. The results show that this type of analysis has excellent results with machine learning algorithms, which makes them very robust and explainable models when it comes to detecting key atoms.
Nowadays, society tends to capture and share videos thanks to the use of social networks or simply the fact of having a device with a camera or software to edit said multimedia file. This makes it difficult to detect the manipulation that has taken place in the file. This work proposes an analysis of multimedia containers to check if their results are comparable to those of a manipulation detection software. The atoms that are the data that make up the container are extracted, to be aware of the trace left by the different programs and then perform a series of experiments with different data sets to obtain the best model. The results show that this type of analysis has excellent results with machine learning algorithms, which makes them very robust and explainable models when it comes to detecting key atoms.
Description
Trabajo de Fin de Grado en Ingeniería Informática, Facultad de Informática UCM, Departamento de de Ingeniería de Software e Inteligencia Artificial, Curso 2022/2023