Label-Efficient Skeleton-based Recognition with Stable-Invertible Graph Convolutional Networks
Hichem Sahbi
TL;DR
This work tackles skeleton-based action recognition under limited labeled data by formulating a principled active-learning framework that designs informative exemplars via an objective balancing representativity, diversity and uncertainty. A stable, invertible graph convolutional network enables mapping data to a latent space with a tractable distribution, enabling efficient exemplar design and faithful reconstruction to the input space. The approach includes a two-block pipeline (display and learning) with two exemplar-design variants (ambient and latent space) and a four-term objective guiding exemplar selection; experiments on two challenging datasets show that the proposed method outperforms baselines and other display strategies, especially in low-label regimes. The proposed stable-invertible GCNs and latent-space sampling have practical implications for scalable skeleton-based recognition in labeling-scarce settings.
Abstract
Skeleton-based action recognition is a hotspot in image processing. A key challenge of this task lies in its dependence on large, manually labeled datasets whose acquisition is costly and time-consuming. This paper devises a novel, label-efficient method for skeleton-based action recognition using graph convolutional networks (GCNs). The contribution of the proposed method resides in learning a novel acquisition function -- scoring the most informative subsets for labeling -- as the optimum of an objective function mixing data representativity, diversity and uncertainty. We also extend this approach by learning the most informative subsets using an invertible GCN which allows mapping data from ambient to latent spaces where the inherent distribution of the data is more easily captured. Extensive experiments, conducted on two challenging skeleton-based recognition datasets, show the effectiveness and the outperformance of our label-frugal GCNs against the related work.