Implementing Adaptive Separable Convolution for Video Frame Interpolation

TL;DR

This paper replicates and adapts the Adaptive Separable Convolution method for video frame interpolation, exploring its performance on smaller datasets and different loss functions to optimize results in data-scarce scenarios.

Contribution

It demonstrates the effectiveness of the adaptive separable convolution approach in low-data settings and evaluates various loss functions for improved video interpolation.

Findings

Visually pleasing videos achieved with the adapted model

Lower evaluation scores compared to original work

Optimal loss functions identified for small datasets

Abstract

As Deep Neural Networks are becoming more popular, much of the attention is being devoted to Computer Vision problems that used to be solved with more traditional approaches. Video frame interpolation is one of such challenges that has seen new research involving various techniques in deep learning. In this paper, we replicate the work of Niklaus et al. on Adaptive Separable Convolution, which claims high quality results on the video frame interpolation task. We apply the same network structure trained on a smaller dataset and experiment with various different loss functions, in order to determine the optimal approach in data-scarce scenarios. The best resulting model is still able to provide visually pleasing videos, although achieving lower evaluation scores.