Semantically Consistent Video Inpainting with Conditional Diffusion Models

arXiv:2405.00251v1 Announce Type: cross
Abstract: Current state-of-the-art methods for video inpainting typically rely on optical flow or attention-based approaches to inpaint masked regions by propagating visual information across frames. While such approaches have led to significant progress on standard benchmarks, they struggle with tasks that require the synthesis of novel content that is not present in other frames. In this paper we reframe video inpainting as a conditional generative modeling problem and present a framework for solving such problems with …

abstract art arxiv attention benchmarks consistent cs.cv cs.lg current diffusion diffusion models flow information inpainting novel optical optical flow progress standard state struggle synthesis tasks type video visual while