Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

MoTIF: Learning Motion Trajectories with Local Implicit Neural Functions for Continuous Space-Time Video Super-Resolution

About

This work addresses continuous space-time video super-resolution (C-STVSR) that aims to up-scale an input video both spatially and temporally by any scaling factors. One key challenge of C-STVSR is to propagate information temporally among the input video frames. To this end, we introduce a space-time local implicit neural function. It has the striking feature of learning forward motion for a continuum of pixels. We motivate the use of forward motion from the perspective of learning individual motion trajectories, as opposed to learning a mixture of motion trajectories with backward motion. To ease motion interpolation, we encode sparsely sampled forward motion extracted from the input video as the contextual input. Along with a reliability-aware splatting and decoding scheme, our framework, termed MoTIF, achieves the state-of-the-art performance on C-STVSR. The source code of MoTIF is available at https://github.com/sichun233746/MoTIF.

Yi-Hsin Chen, Si-Cun Chen, Yi-Hsin Chen, Yen-Yu Lin, Wen-Hsiao Peng• 2023

Related benchmarks

TaskDatasetResultRank
Video Super-ResolutionREDS (val)
PSNR31.03
89
Continuous spatio-temporal video super-resolutionGoPro 85 (out-of-distribution)
PSNR31.53
80
Video Super-ResolutionUDM10 (test)
PSNR24.97
51
Space-Time Video Super-ResolutionVid4 (test)
PSNR25.79
46
Space-Time Video Super-ResolutionGoPro Average (test)
PSNR30.04
45
Space-Time Video Super-ResolutionVid4
PSNR25.79
41
Space-Time Video Super-ResolutionAdobe-Average (test)
PSNR29.82
38
Spatiotemporal Video Super-ResolutionBS-ERGB
PSNR24.21
29
Space-Time Video Super-ResolutionGoPro Center (test)
PSNR31.04
28
Space-Time Video Super-ResolutionAdobe-Center (test)
PSNR30.63
28
Showing 10 of 26 rows

Other info

Follow for update