Video Frame Interpolation via Adaptive Convolution

TL;DR

This paper introduces a novel video frame interpolation method that uses a deep neural network to estimate adaptive convolution kernels, integrating motion estimation and pixel synthesis into a single, end-to-end trainable process for improved handling of occlusion and motion challenges.

Contribution

The proposed method unifies motion estimation and pixel synthesis into a single adaptive convolution process using deep learning, eliminating the need for explicit optical flow and improving interpolation quality.

Findings

Handles occlusion, blur, and brightness changes effectively

Achieves high-quality frame interpolation without optical flow

End-to-end training on standard video data

Abstract

Video frame interpolation typically involves two steps: motion estimation and pixel synthesis. Such a two-step approach heavily depends on the quality of motion estimation. This paper presents a robust video frame interpolation method that combines these two steps into a single process. Specifically, our method considers pixel synthesis for the interpolated frame as local convolution over two input frames. The convolution kernel captures both the local motion between the input frames and the coefficients for pixel synthesis. Our method employs a deep fully convolutional neural network to estimate a spatially-adaptive convolution kernel for each pixel. This deep neural network can be directly trained end to end using widely available video data without any difficult-to-obtain ground-truth data like optical flow. Our experiments show that the formulation of video interpolation as a single convolution process allows our method to gracefully handle challenges like occlusion, blur, and abrupt brightness change and enables high-quality video frame interpolation.