Temporal-Relational CrossTransformers for Few-Shot Action Recognition

TL;DR

This paper introduces Temporal-Relational CrossTransformers (TRX), a novel method for few-shot action recognition that models temporal relations between frames, achieving state-of-the-art results on multiple datasets.

Contribution

The paper presents a new approach using CrossTransformer attention to construct class prototypes from temporal frame tuples, improving few-shot action recognition performance.

Findings

Achieves state-of-the-art results on Kinetics, SSv2, HMDB51, UCF101

Outperforms prior work on SSv2 by 12%

Highlights the importance of modeling temporal relations

Abstract

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.