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

We investigate video transforms that result in class-homogeneouslabel-transforms. These are video transforms that consistently maintain ormodify the labels of all videos in each class. We propose a general approach todiscover invariant classes, whose transformed examples maintain their label;pairs of equivariant classes, whose transformed examples exchange their labels;and novel-generating classes, whose transformed examples belong to a new classoutside the dataset. Label transforms offer additional supervision previouslyunexplored in video recognition benefiting data augmentation and enablingzero-shot learning opportunities by learning a class from transformed videos ofits counterpart.  Amongst such video transforms, we study horizontal-flipping, time-reversal,and their composition. We highlight errors in naively using horizontal-flippingas a form of data augmentation in video. Next, we validate the realism oftime-reversed videos through a human perception study where people exhibitequal preference for forward and time-reversed videos. Finally, we test ourapproach on two datasets, Jester and Something-Something, evaluating the threevideo transforms for zero-shot learning and data augmentation. Our results showthat gestures such as zooming in can be learnt from zooming out in a zero-shotsetting, as well as more complex actions with state transitions such as diggingsomething out of something from burying something in something.

Retro-Actions: Learning 'Close' by Time-Reversing 'Open' Videos