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 address the problem of data augmentation for video action recognition.Standard augmentation strategies in video are hand-designed and sample thespace of possible augmented data points either at random, without knowing whichaugmented points will be better, or through heuristics. We propose to learnwhat makes a good video for action recognition and select only high-qualitysamples for augmentation. In particular, we choose video compositing of aforeground and a background video as the data augmentation process, whichresults in diverse and realistic new samples. We learn which pairs of videos toaugment without having to actually composite them. This reduces the space ofpossible augmentations, which has two advantages: it saves computational costand increases the accuracy of the final trained classifier, as the augmentedpairs are of higher quality than average. We present experimental results onthe entire spectrum of training settings: few-shot, semi-supervised and fullysupervised. We observe consistent improvements across all of them over priorwork and baselines on Kinetics, UCF101, HMDB51, and achieve a newstate-of-the-art on settings with limited data. We see improvements of up to8.6% in the semi-supervised setting.

Learn2Augment: Learning to Composite Videos for Data Augmentation in Action Recognition