On the Utility of 3D Hand Poses for Action Recognition
This work addresses action recognition, particularly for applications with limited compute budgets, by efficiently leveraging 3D hand poses, though it is incremental as it builds on existing multimodal and transformer-based approaches.
The paper tackled the problem of action recognition by proposing HandFormer, a multimodal transformer that combines 3D hand poses and sparse RGB frames to model hand-object interactions, achieving new state-of-the-art performance on Assembly101 and H2O datasets with significant improvements in egocentric action recognition.
3D hand pose is an underexplored modality for action recognition. Poses are compact yet informative and can greatly benefit applications with limited compute budgets. However, poses alone offer an incomplete understanding of actions, as they cannot fully capture objects and environments with which humans interact. We propose HandFormer, a novel multimodal transformer, to efficiently model hand-object interactions. HandFormer combines 3D hand poses at a high temporal resolution for fine-grained motion modeling with sparsely sampled RGB frames for encoding scene semantics. Observing the unique characteristics of hand poses, we temporally factorize hand modeling and represent each joint by its short-term trajectories. This factorized pose representation combined with sparse RGB samples is remarkably efficient and highly accurate. Unimodal HandFormer with only hand poses outperforms existing skeleton-based methods at 5x fewer FLOPs. With RGB, we achieve new state-of-the-art performance on Assembly101 and H2O with significant improvements in egocentric action recognition.