Temporal copying and local hallucination for video Inpainting

Video inpainting is the task of removing objects from videos. In particular, the goal is not only to fill every frame with plausible content but also to maintain a temporal consistency so that no abrupt changes can be perceived. The current state of the art in video inpainting, which builds upon dee...

Descripción completa

Detalles Bibliográficos
Autores: álvarez De La Torre, David, Álvarez de la Torre, David
Tipo de recurso: tesis de maestría
Fecha de publicación:2020
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/334882
Acceso en línea:https://hdl.handle.net/2117/334882
Access Level:acceso abierto
Palabra clave:Machine learning
Neural networks (Computer science)
Video recording
Computer vision
deep learning
computer vision
video inpainting
aprendizaje automático
aprendizaje profundo
redes neuronales
visión por computador
Aprenentatge automàtic
Xarxes neuronals (Informàtica)
Vídeo
Visió per ordinador
Àrees temàtiques de la UPC::Enginyeria de la telecomunicació
Descripción
Sumario:Video inpainting is the task of removing objects from videos. In particular, the goal is not only to fill every frame with plausible content but also to maintain a temporal consistency so that no abrupt changes can be perceived. The current state of the art in video inpainting, which builds upon deep neural network, suffers from the problem of handling large amounts of frames when working with decent resolution frames. In our work, we propose to tackle the problem of video inpainting by dividing it into two independent sub-tasks. The first, a Dense Flow Prediction Network (DFPN) capable of predicting the movement of the background by taking into account the movement of the object to remove. And the second, a Copy-and-Hallucinate Network (CHN) that uses the output of the previous network to copy the regions that are visible in reference frames while hallucinating those that are not. Both networks are trained independently and mixed using one of our three algorithm proposals: the Frame-by-Frame (FF) algorithm, the Inpaint-and-Propagate (IP) algorithm or the Copy-and-Propagate (CP) algorithm. We analyze our results by taking both an objective and a subjective approach in two different data sets. In both cases, we realize that our models are close to the state of the art but do not overpass it.