Efficient Space-time Video Super Resolution using Low-Resolution Flow and Mask Upsampling

TL;DR

This paper presents an efficient method for space-time video super-resolution that combines low-resolution flow interpolation, super-resolution, and mask upsampling, resulting in improved quality and efficiency over existing methods.

Contribution

It introduces a novel, lightweight approach that reuses flow and mask information in high-resolution space, reducing artifacts and improving performance in space-time super-resolution.

Findings

Outperforms current state-of-the-art models on REDS STSR validation set

Reduces memory usage and inference time compared to sequential methods

Improves intermediate frame quality with a residual refinement network

Abstract

This paper explores an efficient solution for Space-time Super-Resolution, aiming to generate High-resolution Slow-motion videos from Low Resolution and Low Frame rate videos. A simplistic solution is the sequential running of Video Super Resolution and Video Frame interpolation models. However, this type of solutions are memory inefficient, have high inference time, and could not make the proper use of space-time relation property. To this extent, we first interpolate in LR space using quadratic modeling. Input LR frames are super-resolved using a state-of-the-art Video Super-Resolution method. Flowmaps and blending mask which are used to synthesize LR interpolated frame is reused in HR space using bilinear upsampling. This leads to a coarse estimate of HR intermediate frame which often contains artifacts along motion boundaries. We use a refinement network to improve the quality of HR intermediate frame via residual learning. Our model is lightweight and performs better than current state-of-the-art models in REDS STSR Validation set.