notesum.ai

Published at December 10

Repurposing Pre-trained Video Diffusion Models for Event-based Video Interpolation

cs.CV

Released Date: December 10, 2024

Authors: Jingxi Chen¹, Brandon Y. Feng², Haoming Cai, Tianfu Wang, Levi Burner, Dehao Yuan, Cornelia Fermuller, Christopher A. Metzler, Yiannis Aloimonos

Aff.: ¹University of Maryland, College Park; ²Massachusetts Institute of Technology

Arxiv: http://arxiv.org/pdf/2412.07761v1

Feature	Video Generation	EVFI
Dataset Size [7, 29, 20]	$10^{7}$	$10^{2}$
Dataset Quality [7, 29, 20, 18, 38, 35]	Commercial	Custom
Model Type [7, 38, 36, 23, 18]	Video Diffusion (Foundation) Models	Custom Models
Model Size [7, 29, 38, 36, 23, 18]	$10^{10}$	$10^{8}$
Video Generation Type [7, 38, 36, 23, 18]	Extrapolation	Interpolation