Deep Iterative Frame Interpolation for Full-frame Video Stabilization

TL;DR

This paper introduces an unsupervised deep learning framework for full-frame video stabilization that avoids cropping and distortion by using iterative frame interpolation, achieving near real-time performance.

Contribution

It presents the first unsupervised deep approach to full-frame video stabilization that maintains frame integrity without cropping or distortion.

Findings

Reduces inter-frame jitter effectively

Operates in near real-time at 15 fps

Outperforms state-of-the-art methods in quality

Abstract

Video stabilization is a fundamental and important technique for higher quality videos. Prior works have extensively explored video stabilization, but most of them involve cropping of the frame boundaries and introduce moderate levels of distortion. We present a novel deep approach to video stabilization which can generate video frames without cropping and low distortion. The proposed framework utilizes frame interpolation techniques to generate in between frames, leading to reduced inter-frame jitter. Once applied in an iterative fashion, the stabilization effect becomes stronger. A major advantage is that our framework is end-to-end trainable in an unsupervised manner. In addition, our method is able to run in near real-time (15 fps). To the best of our knowledge, this is the first work to propose an unsupervised deep approach to full-frame video stabilization. We show the advantages of our method through quantitative and qualitative evaluations comparing to the state-of-the-art methods.