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

Although vital to computer vision systems, few-shot action recognition isstill not mature despite the wide research of few-shot image classification.Popular few-shot learning algorithms extract a transferable embedding from seenclasses and reuse it on unseen classes by constructing a metric-basedclassifier. One main obstacle to applying these algorithms in actionrecognition is the complex structure of videos. Some existing solutions sampleframes from a video and aggregate their embeddings to form a video-levelrepresentation, neglecting important temporal relations. Others perform anexplicit sequence matching between two videos and define their distance asmatching cost, imposing too strong restrictions on sequence ordering. In thispaper, we propose Compromised Metric via Optimal Transport (CMOT) to combinethe advantages of these two solutions. CMOT simultaneously considers semanticand temporal information in videos under Optimal Transport framework, and isdiscriminative for both content-sensitive and ordering-sensitive tasks. Indetail, given two videos, we sample segments from them and cast the calculationof their distance as an optimal transport problem between two segmentsequences. To preserve the inherent temporal ordering information, weadditionally amend the ground cost matrix by penalizing it with the positionaldistance between a pair of segments. Empirical results on benchmark datasetsdemonstrate the superiority of CMOT.

Few-Shot Action Recognition with Compromised Metric via Optimal Transport