English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

When a very fast dynamic event is recorded with a low-framerate camera, theresulting video suffers from severe motion blur (due to exposure time) andmotion aliasing (due to low sampling rate in time). True TemporalSuper-Resolution (TSR) is more than just Temporal-Interpolation (increasingframerate). It can also recover new high temporal frequencies beyond thetemporal Nyquist limit of the input video, thus resolving both motion-blur andmotion-aliasing effects that temporal frame interpolation (as sophisticated asit maybe) cannot undo. In this paper we propose a "Deep Internal Learning"approach for true TSR. We train a video-specific CNN on examples extracteddirectly from the low-framerate input video. Our method exploits the strongrecurrence of small space-time patches inside a single video sequence, bothwithin and across different spatio-temporal scales of the video. We furtherobserve (for the first time) that small space-time patches recur alsoacross-dimensions of the video sequence - i.e., by swapping the spatial andtemporal dimensions. In particular, the higher spatial resolution of videoframes provides strong examples as to how to increase the temporal resolutionof that video. Such internal video-specific examples give rise to strongself-supervision, requiring no data but the input video itself. This results inZero-Shot Temporal-SR of complex videos, which removes both motion blur andmotion aliasing, outperforming previous supervised methods trained on externalvideo datasets.

Across Scales & Across Dimensions: Temporal Super-Resolution using Deep Internal Learning