Active Learning for GCN-based Action Recognition

TL;DR

This paper introduces a label-efficient GCN model for skeleton-based action recognition, utilizing an adversarial acquisition function and bidirectional architectures to improve performance with fewer labeled samples.

Contribution

It proposes a novel adversarial acquisition function and bidirectional GCN architectures to enhance label efficiency in action recognition.

Findings

Significant performance improvements over prior methods.

Effective identification of informative exemplars for labeling.

Enhanced understanding of data distribution through bidirectional GCNs.

Abstract

Despite the notable success of graph convolutional networks (GCNs) in skeleton-based action recognition, their performance often depends on large volumes of labeled data, which are frequently scarce in practical settings. To address this limitation, we propose a novel label-efficient GCN model. Our work makes two primary contributions. First, we develop a novel acquisition function that employs an adversarial strategy to identify a compact set of informative exemplars for labeling. This selection process balances representativeness, diversity, and uncertainty. Second, we introduce bidirectional and stable GCN architectures. These enhanced networks facilitate a more effective mapping between the ambient and latent data spaces, enabling a better understanding of the learned exemplar distribution. Extensive evaluations on two challenging skeleton-based action recognition benchmarks reveal significant improvements achieved by our label-efficient GCNs compared to prior work.