Temporal-Relational CrossTransformers for Few-Shot Action Recognition

Research output: Contribution to conferenceConference Paperpeer-review

135 Downloads (Pure)


We propose a novel approach to few-shot action recognition, finding temporally-corresponding frame tuples between the query and videos in the support set. Distinct from previous few-shot works, we construct class prototypes using the CrossTransformer attention mechanism to observe relevant sub-sequences of all support videos, rather than using class averages or single best matches. Video representations are formed from ordered tuples of varying numbers of frames, which allows sub-sequences of actions at different speeds and temporal offsets to be compared.

Our proposed Temporal-Relational CrossTransformers (TRX) achieve state-of-the-art results on few-shot splits of Kinetics, Something-Something V2 (SSv2), HMDB51 and UCF101. Importantly, our method outperforms prior work on SSv2 by a wide margin (12%) due to the its ability to model temporal relations. A detailed ablation showcases the importance of matching to multiple support set videos and learning higher-order relational CrossTransformers.
Original languageEnglish
Number of pages8
Publication statusPublished - 25 Jun 2021
EventComputer Vision and Pattern Recognition 2021 - Online
Duration: 19 Jun 202125 Jun 2021


ConferenceComputer Vision and Pattern Recognition 2021
Abbreviated titleCVPR
Internet address

Structured keywords

  • Digital Health


Dive into the research topics of 'Temporal-Relational CrossTransformers for Few-Shot Action Recognition'. Together they form a unique fingerprint.

Cite this